Identification system in addition to video surveillance systems. The algorithm will search for evidence. Biometric identification in integrated security systems Person identification system

Everyone knows the scenes from science fiction films: the hero approaches the door and the door responds, having recognized him. This is one of the primary demonstrations of the usability and reliability of biometric technologies for access control. However, in practice it is not so simple. Today, companies are ready to introduce access control to their employees using biometric technologies.

Traditional methods of identifying an individual, which are based on various identification cards, keys or unique data, such as, for example, a password, are not reliable in this world, which is required today's day. A natural way to improve the reliability of identifiers was the use of biometric technologies for security systems.

The range of problems that can be found using new technologies is extremely wide:

1. prevent the penetration of criminals on the territory and in the premises for the storage of details, theft of documents, cards, passwords;
2. restrict access to information and ensure personal reliability for saving it;
3. ensure access to third-party facilities without certified specialists;
4. eliminating overhead costs associated with the operation of access control systems (cards, keys);
5. turn off inconveniences associated with wasting, wasting or simply forgetting keys, cards, passwords;
6. organize the area of ​​access and availability of satellites.

The development of sophisticated technologies for recognizing images from various biometric characteristics began long ago; the beginnings were laid in the 60s. Our spy scientists have achieved significant success in the development of theoretical ambushes of these technologies. Prote practical results are taken away from today’s advances and from “yesterday”. The sophistication of modern computers and sophisticated algorithms have made it possible to create products that, due to their characteristics and characteristics, have become accessible and useful to a wide range of traders.

The idea of ​​using a person's individual characteristics to identify them is not new. Today there are low technologies that can be used in security systems to identify an individual for:

1. the fingers are broken (either the fingers are stiff or the hands are shaved);
2. rice (based on optical and infrared images);
3. iris of the eyes;
4. voice;
5. Other characteristics.

All biometric technologies have advanced approaches to the ultimate task of identification, although all methods are subject to ease of operation and accuracy of results.

Whether biometric technology is being implemented step by step:

1. scanning the object;
2. obtaining individual information;
3. molding a template;
4. Alignment of the flow template with the database.

The biometric recognition system establishes the correspondence of specific physiological and behavioral characteristics of the patient to a given pattern. Therefore, the biometric system consists of two modules: a registration module and an identification module.

Registration module"begins" a system to identify a specific person. At the registration stage, the video camera or other sensors scan the person in order to create a digital image. The scanning process takes approximately 20 - 30 seconds, resulting in the formation of an image. Ideally, these images have a few different angles and expressions to allow for more accurate data to be captured. A special software module collects the phenomenon and identifies the characteristic features, then creates a template. There are certain parts of the individual that practically do not change over time, for example, the upper contours of the eye pits, areas of the upper part of the face, and the edges of the mouth. Most algorithms developed for biometric technologies make it possible to make possible changes in the human population, without being necessary to analyze the area of ​​facial hair between hair growth. The image template of the skin peeler is saved in the biometric system database.

Identification module captures a person's image from a video camera and converts it to the same digital format in which the template is saved. Once the data is removed, check the template that is stored in the database in order to determine how each image corresponds to one another. The level of similarity required for verification is a further level that can be adjusted for different types of personnel, PC severity, hours of production and a number of other factors.

Identification can be in the form of verification, authentication or recognition. During verification, the identity of the extracted data and the template that is saved in the database is confirmed. Authentication - confirms the identity of the image contained in the video camera to one of the templates that are stored in the database. When recognized, once the characteristics of one of the templates that are saved are found to be the same, the system identifies a person with a similar template.

When using biometric systems, especially facial recognition systems, it is important to ensure that the correct biometric characteristics are introduced so that authentication decisions are made correctly. However, there are few particularities and, first of all, about those that many biometric characteristics can be changed. It is clear that the system is incredibly reliable. Moreover, with the use of different technologies, the mixture can be completely reduced. For access control systems, with the rise of biometric technologies, it is necessary to understand that it is more important not to let in “someone else” than to let in all “our own”.

An important factor for the use of biometric technologies in security systems is the ease of use. A person whose characteristics are scanned is not guilty of recognizing everyday incompetence. In this case, the best method is the technology of recognition by appearance. However, there are other problems associated with the accuracy of the system.

Regardless of the obvious advantages, there are few negative advances against biometrics, which are often called out by those who will not use biometric data to punish people and violate their right to privacy. Through sensational statements and unreported news, the adoption of biometric technologies is sharply increasing in relation to the real world.

And yet, the use of biometric identification methods has become particularly relevant in the future. This problem became especially acute after the 11th of Sunday in the United States. The world's intelligence realized the stage of growing threat of terrorism in the current world and the complexity of organizing reliable defense using traditional methods. These very tragic events have become a starting point for increasing respect for the current integration of security systems. The official thought is that if control at airports were in place, then the misfortune could have ended. These days, the sounds of people at low risk may lead to significant relief under the hour of the rise of modern video surveillance systems in integration with person recognition systems.

At this time, there are several main methods for recognizing an individual:

1. "eigenfaces";
2. analysis of "sweet rice";
3. analysis based on “neural measurements”;
4. method of "automatic processing of revealing images".

All these methods are subject to complexity of implementation and method of hardening.

"Eigenface" can be translated as "vlasne oblichya". This technology creates two-dimensional images in shades of gray, which become important characteristics of the image. The "eigenface" method is often cited as a basis for other methods of detection.

Combination characteristics 100 - 120 "eigenface" can be updated in a large number of ways. At the time of registration, the “eigenface” of a specific person’s skin appears to a number of coefficients. For the authentication mode, in which images are checked to verify identity, the “live” template is equal to the registered template, according to the coefficient of validity. The level of difference between the templates and indicates the fact of identification. The "eigenface" technology is optimal for choosing well-lit areas, if it is possible to scan the face at the front.

The technique for analyzing “extraordinary rice” is the identification technology that is most widely used. This technology is similar to the “Eigenface” technique, but in a larger world it is adapted to change the appearance or facial expressions of a person (smiling or frowning). The technology of “extraordinary rice” has dozens of characteristic features of different areas of the individual, and with the harmonization of its remarkable growth. The individual combination of these parameters determines the characteristics of the skin of a particular individual. A person is unique, but also dynamic, because... People can laugh, grow a beard, wear out their eyepieces - all this makes the identification procedure more complicated. In this way, for example, when laughing, one is careful not to displace parts of the individual, which have been spread out throughout the whole mouth, so that in one’s own body we can call a similar mix of smaller parts. Doctors with such assumptions can clearly identify a person and expose him for various mimetic changes. Since this analysis looks at local areas of an individual, acceptable treatments can range up to 25° at the horizontal plane, and up to approximately 15° at the vertical plane, and will require quite laborious and expensive equipment, The stage of expansion of this method is obviously speeding up.

The method, based on neural interconnection, has the characteristic features of both individuals - the registered one is verified to be successful. “Neural measurements” use an algorithm that establishes the unique parameters of a person that are verified, and the parameters of a template that is in the database, which sets the maximum possible number of parameters. Inconsistencies are regularly identified and the template is checked against the database, then a mechanism is launched that, with the help of various types of coefficients, determines the stage of The identity of the individual that is checked against the template in the database. This method is more accurate in identifying an individual in capable minds.

The method of “automatic processing of the facial image” is the simplest technology that vikorista and the relationship between facial points that are easily identified, such as the eyes, the end of the nose, the corners of the mouth. Although this method is not so tight as “eigenfaces” or “neural networking”, you can achieve effective results in the minds of weak illumination.

Identification systems present on the market

To date, a small number of commercial products have been developed that are used for the recognition of individuals. The algorithms that are used in these products vary and it is still difficult to assess how the technology has advantages. The leaders are the following systems: Visionic, Viisage and Miros.

• Visionic's FaceIt software is based on an algorithm for analyzing local signs at Rockefeller University. One commercial company in Great Britain integrated FaceIt into a television anti-crime system called Mandrake. This system searches for criminals for video data from 144 cameras located in a closed area. If an identity is established, the system notifies the security officer about the identity. In Russia, the representative of the Visionic company is the DanCom company.
• Another leader in this field, the company Viisage, is developing an algorithm developed by the Massachusetts Institute of Technology. Commercial companies and government structures in rich American states and in other countries use the Visage company system along with identification certificates, for example, water rights.
• ZN Vision Technologies AG (Nimechina) introduces to the market a low number of products that lack feature recognition technology. These systems are presented to the Russian market by the Soling company.
• The TrueFace personality recognition system of the Miros company uses neural tracking technology, and the system itself is part of a complex produced by the Mr. Payroll corporation and installed at casinos and other important depositories in wealthy US states.

Independent experts carried out extensive testing of various character recognition technologies. The test results are shown below.

Rice. 1. Rough analysis of the effectiveness of feature recognition in different systems

In practice, with the use of personality recognition systems in stock of standard electronic security systems, it is transmitted that people, as soon as they are identified, look directly at the camera. Thus, the system works with very simple two-dimensional images, which significantly simplifies the algorithms and reduces the intensity of calculations. However, in this case, recognition is still not trivial, since the algorithms are responsible for the ability to change the level of illumination, change the appearance of the face, visibility or the presence of makeup more eyepieces

The reliability of the robotic recognition system depends on several factors:

• Brightness of the image. There is a noticeable decrease in the reliability of the silent robotic system, since the people we are trying to identify are not looking directly at the camera, but are being filmed in poor lighting.
• The relevance of the photograph entered into the database.
• Database size.

Recognition technologies work well with standard video cameras that transmit data and are stored on a personal computer, and achieve a resolution of 320×240 pixels per inch with a video stream speed of 3 – 5 frames per second. To be fair, a good video conferencing solution will require a video stream speed of at least 15 frames per second. High-quality video streams with a high level of separation lead to improved identification capabilities. When recognizing individuals from a large distance, there is a strong correlation between the video camera and the result of identification.

A total of 10,000 images are collected from databases using standard personal computers.

Visnovok

The current methods for recognizing the characteristics of the plant are close to being widely used, but it is still impossible to trust the opening of doors without the technology of recognition by appearance. Vaughn is the assistant for the protection of other access control systems.

This method itself is used in many situations, if it is necessary to ensure that the presentation of the document is effectively due to the person who presented it. This happens, for example, at an international airport, when the guard checks the passport photo of the person holding the passport and makes a decision on which passport to take. A similar algorithm can be accessed by a computer system. The significance lies only in the fact that the photograph matches the template that is already stored in the database.

Technologies have already appeared that rely on recognized features in infrared light. The new technology is based on the fact that a thermal image created by the transfer of heat by blood vessels, or, in other words, a thermogram of a person, is unique to the skin and, perhaps, as a biometric characteristic of access control systems. This thermogram is a stable identifier, the lower geometry of the face, which may be subject to changes in a person’s appearance.

Currently, biometrics is increasingly penetrating our lives. The leading edges of the world have already announced and are planning to issue electronic passports in the near future, which will provide information about the biometric characteristics of their holder; Many office centers have introduced biometric sensors in corporate access control systems; laptops have long been equipped with biometric customer authentication; The established security services are constantly able to detect any kind of villain that is heard in a crowd of people

Andriy Khrulov
Head of the Biometric Department
and comprehensive security systems
Group of companies "Technoserv", Ph.D.

There are more applications for biometric systems. The success of biometrics is easy to explain. Traditional methods of identifying identity, based on the principles “I am what I know” (identification cards, tokens, documents that can be identified) and “I am what I know” (passwords, PIN codes) - not fully i. The card can be easily wasted, the password can be forgotten, and even if there is a malicious person, the system will not be able to identify you as a fake person.

In addition, traditional methods of identification are absolutely marginal when it comes to specifying the required identification feature, and there are more such tasks:

• recognize the villain from the crowd;
• check that the passport is presented by your ruler;
• find out what people are doing when they are having fun;
• understand that the person has not previously been involved in financial fraud involving loans;
• identify potentially dangerous patients before entering the stadium.

All these tasks can be achieved only with the help of various biometric identification features that are based on the principle “I am who I am”. This principle allows the information system to identify directly the person, and not the objects it provides, but the information it reports.

The uniqueness of biometrics

There is a wide variety of biometric characteristics of a person that are used to identify their particularity, which means the image of their appearance. Biometrics is unique in that it does not require special sensors to capture images - the images can be captured using a standard video security camera. Moreover, a photograph of an individual is practically present on any document that identifies the individual, and therefore, the advancement of this technology is in practice not related to various regulatory problems and the complexities of the social adoption of technology.

This also means that facial images can be captured indirectly for the person themselves, and therefore, facial biometrics are optimally suited for monitoring and identification systems.

Any system for recognizing an individual is a typical system for recognizing images, which comes down to the formation of a certain set of signs, the so-called biometric template, similar to the underlying system of a mathematical model. This model itself becomes the key know-how of any biometric system, and the effectiveness of recognition depends largely on such factors as the stability of the biometric pattern to various transients, due to the output photo Images.

The effectiveness of identification recognition depends entirely on such factors as the resistance of the biometric template to various changes in the output photo or video image.

Regardless of the diversity of individual recognition systems presented both on the Russian market and in the world, many of them use the same biometric engines - based on software implementation of methods and the updating of mathematical models of an individual. Russia has the greatest expansion of such biometric engines as Cognitec (developed by Cognitec Systems GmbH, Germany), Cascade-Potik (developed by Technoserv, Russia), FRS SDK (developed by Asia Software, Kazakhstan), FaceIt company L1 Identity Solutions, USA).

As a rule, recognition of a person in any biometric engine consists of a number of stages: identification of an individual, assessment of personality, identification of a pattern, creation and making of a decision.

Stage 1: exposure

At this stage, the system automatically sees (detects) people in a series of video frames or photographs, and the range of angles and scales of features can significantly change, which is extremely important for alerting security systems. It is completely unnecessary that all seen individuals be identified (as a rule, this is not possible), but to identify the maximum number of individuals in the future and, if necessary, place them in archives around the edges (Fig. 1).

Identification of an individual is one of the key stages of recognition, so skipping detection by the detector automatically means the impossibility of further identification. The brightness of the detector is usually characterized by the intensity of detection of P0. For current biometric systems that operate in the minds of a stream of people, the significance of the revealed detection rate becomes 95 to 99% and lies in the minds of video recording (lighting, separate cameras, etc.).

One of the most promising trends in the development of the biometrics market is the emergence of intelligent digital video cameras that implement the function of identifying the structure of the built-in logic (Fig. 2). Smart video cameras allow you to capture a clear video stream, and the metadata associated with it, to provide information about the detected features.

This approach makes it possible to significantly reduce the hardware requirements of the recognition system, which in turn reduces the end-usability of biometric complexes, making them more accessible to the end user. In addition, the possibilities for data transmission channels will change, since this approach does not require gigabit lines to transmit high-quality video, but rather the availability of standard measures for transmitting compressed video and an insignificant stream of detected images.

Stage 2: juiciness assessment

This is a very important stage of recognition, in which the biometric engine selects from a large array of detected individuals the most sensitive images that meet the specified criteria of accuracy.

Developers of biometric systems often lie, insisting that their system will ensure a high level of recognition, consistent with the image of an individual in a video stream based on the accuracy specified in GOST R ISO/IEC 19794-5. However, this GOST imposes even harsh (almost ideal) restrictions on the brightness of photographs of an individual (frontal angle of an individual with a view of no more than 5 degrees; uniform lightening; neutral facial expressions, but not may be in the real minds of video surveillance systems. Such benefits of DSTU are fully justified by the fact that this standard is intended to unify the format of saving electronic photographs in passport and visa documents of the new generation - the so-called biometric passports. In practice, biometric identification systems for the detection of anxiety in mothers work with significantly less sympathetic minds:

• recovery of the individual from the frontal position on the slope, which exceeds 20 degrees;
• more overexposed;
• perekrittya parts of an individual;
• the presence of shadows on the surface;
• minimum image size, etc.

The very stability of the biometric engine in such folding minds means its strength. In current biometric engines, at the stage of capacity assessment, the following are usually assessed:

• angle of the individual (not guilty of exaggerating 20-30 degrees);
• the size of the individual (estimated by the distance between the eyes and should be more than 50–80 px);
• Partially closed individuals (closed individuals are not to blame for 10–25% of the individual's occult area).

There is a hidden truth that if the person’s eyes are closed in the image (by blinking or eyepieces), then the system will not be able to recognize the person. Indeed, early face recognition algorithms used the centers of the eyes as a basis for further image processing, as well as for standard face scaling. However, many modern biometric engines (for example, Cognitec or “Kaskad-Potik”) use more complex coding schemes and are not tied to the position of the centers of the points.

Stage 3: depending on the template

This is one of the most complex and unique stages of identification recognition, which becomes the key know-how of biometric engine technology. The essence of this stage lies in the non-trivial mathematical transformation of the image into a set of characters, combined with a biometric template.

The skin of the individual resembles the moisture and unique biometric pattern. The principles of biometric patterns are extremely varied: a pattern can be based on textural features of the face, geometric features, characteristic points, combinations of different different characters.

The most important characteristic of a biometric template is its size. The larger the template, the more informative symbols it includes, and the lower the fluidity and efficiency of the template. Typical template sizes for biometric systems range from 1 to 20 kbytes.

Stage 4: making and making decisions

This is the combined stage of the recognition system, in which the biometric template of the individual identified by the detected individual is compared with the array of templates that are stored in the database. In its simplest form, the process consists of a simple enumeration of all the templates and an assessment of their similarities. On the basis of the evaluations and their comparison with the specified thresholds, decisions are made about the presence or presence of an identical individual in the database.

In current systems, the alignment is implemented using complex optimal alignment schemes, which ensure alignment speed of 10,000 to 200,000 alignments per second or more. Moreover, it is important to understand that the creation process can be parallel, which allows identification systems to practically in real time create images of, for example, 100,000 people behind large arrays.

The effectiveness of personal recognition systems is usually characterized by authenticity of identification. Obviously, biometric identification can result in two types of rewards.

1. The first mercy is associated with the possibility of missing and not recognizing a person who is actually in the database - they are often called the first mercy. Moreover, they often indicate not the significance of the mercy of the first family, but one minus the importance of the mercy of the first family. This is called the importance of correct recognition of PPR.
2. Another kind of mercy displays episodes when the system recognizes a person who is not actually in the database or is confused with another person - they are usually called a kind of mercy of a different kind. For current systems of individual recognition, the typical values ​​of the probability of correct recognition are usually in the range of 80 to 97%; for animals of a different kind it does not exceed 1%.

Successful identification

Varto understand that detection recognition is not an absolute technology. One can often criticize biometric systems in that real objects do not achieve the same high performance as in “laboratory” systems. This assertion is more correct than ever. In truth, it is only possible to effectively recognize the evidence from the best minds, and it is extremely important that before the introduction of biometrics, individuals understand which minds operate the system. However, for most current systems, recognition of the mind is entirely possible on real objects. Thus, to increase the effectiveness of individual recognition in identification zones, organize a direct flow of people (door openers, metal detector frames, turnstiles, etc.) to ensure short-term fixation (no more than 1–2 s) ї individuals of the skin fluid. In this case, the video recording cameras must be installed in such a way that they capture the recorded images from the front position without exceeding 20-30 degrees. (for example, installing cameras at a distance from the aisle zone of 8-10 m with a suspension height of 2-3 m).

The development of these minds with the introduction of recognition systems makes it possible to effectively identify the specific characteristics and characteristics of people who are of interest, with a probability that is as close as possible to the values ​​of the indicators declared by retailers. personal identification.

1

Everywhere, the introduction of video surveillance systems led to the emergence of recognition and identification systems based on video recording. Such systems are based on technology that uses computer surveillance methods to automatically extract various information on a platform for analyzing the sequence of images that come from video cameras in real time. In other words, this technology is based on methods and algorithms for image recognition and image processing as a result of video stream analysis. Daily recognition and identification systems for video recording of buildings without the participation of people can detect and identify in real time a given target (for example, a car, a group of people) or are potentially unsafe with situations (for example, smoke, occupancy, unauthorized access), after which an alarm signal is promptly generated. Another advantage of the stagnation of such systems is the reduction in demand on communication channels and the archive database for filtering video streams in real time. Any recognition and identification system for video recording is based on a unique algorithm for identifying the similarity of the read characteristics to a given pattern. The video camera transmits the video stream to the server in real time, the recognition and identification system determines the reliability of the information, which is stored in the database, and identification is confirmed with arranging behind the importance of the system of factors (for example, weights or headdresses).

recognition system

Identification from video recording.

1. Brilyuk D.V., Starovoitov V.V. Recognizing the people behind the images using neural methods. - Minsk, 2002. - 54 p. (Preprint/Institute of Technical Cybernetics of the National Academy of Sciences of Belarus; No. 2).

2. Kulyabichov Yu.P., Pivtoratska S.V. Review of methods for identifying people based on the characteristics of visual recognition

3. Rogozin O.V., Kladov S.A. Routine analysis of face recognition algorithms for the task of visual identification MDTU im. Not. Bauman, Moscow, 105005, Russia

4. Sherstobitov A.I., Fedosov V.P., Prikhodchenko V.A., Timofeev D.V. Recognition of faces in group photographs using different segmentation algorithms

5. Face Recognition by Elastic Bunch Graph Matching Laurenz Wiskott, Jean-Marc Fellous, Norbert Kruger, and Christoph von der Malsburg,

Enter

Today, modern video surveillance systems implement functionality not only for recording the video stream and displaying the image on the screen, but also for providing various analytical functions. As a rule, such functions belong to the automation of video information analysis. One of the most required functions is the recognition and identification of features in the control zone. In such systems, as physical parameters, the key features of the human individual are determined, which can be distinguished from many others.

Main part

It is significant that there is a shift in the order that is detected by the recognition and identification systems of persons from video recording in real time:

1. Verification. The person recognition and identification system with real-time video recording can confirm a person by matching the standard character recorded in the system earlier. In fact, there is a one-to-one match between the schemes.

2. Identification on a closed multiplier. The system of recognition and identification of persons with real-time video captures images from the video stream without recording the patterns of different people to identify the characteristics that ok. This task can be accomplished by ensuring that the person whose image is assigned is in the database. This task is based on the alignment of characters according to the “one to many” scheme.

The scope of stagnation of such systems in the face of advancing floods is significant:

1. Recognition of the characteristics of thieves and criminals in large towns or at great places. Recognition and identification systems for video recording capture real-time data from video surveillance systems of various government entities and cameras installed in the area. On the basis of the stolen information, the search for thieves and criminals is carried out with the method of further transferring the data to law enforcement agencies.

Today, great projects are being launched in Russia aimed at recognizing and identifying video recording objects in real time. Since spring 2017, a portion of 170,000 video surveillance cameras in Moscow were connected to the fingerprint recognition system. The main monitoring system includes video cameras, cameras on the territory and in school buildings and kindergartens, stadiums, public transport stations and bus stations, near parks, underground passages and other urban areas. This approach has now become an additional tool in the search for evildoers and violators.

Recognizing the personalities of people around the world gives law enforcement officers unique capabilities. Access to the local security system was denied for about 16 thousand. Specialists of law enforcement agencies, state and municipal organizations. The system differentiates access for different groups of clients, which allows you to preserve the confidentiality of the actions of residents of the area.

2. Security of access control. In this type, video security is integrated into the security system and is monitored by controllers on the turnstiles.

Such a system can be set up as both a main one (which makes decisions about allowing a subject to enter a closed territory) and a backup one. Invisible guards will not deny access to the restricted area, and their photos will be saved in the database for the purpose of further processing of the incident by the security service.

Therefore, such systems are installed in large enterprises to ensure the efficiency of the company’s work (for example, the development of advanced technologies). The system automatically recognizes all active military personnel and matches the database. In case of uncertainty or absence of a person in the system, the system activates security protocols and notifies the security services.

The advantage of such stagnation is the minimization of human participation in the process, the enhancement of labor discipline of labor force workers and the reduction of expenses of government entities to pay for work.

3. Anti-theft from hypermarkets and shopping centers. The problem of systematic theft faces every great store with great trading areas. The truth is that existing video surveillance systems are not effective in stopping theft and require theft to be used only to capture evidence of theft, if it has already been detected. Personnel recognition and identification systems with video recording will detect repeated potential thefts in cases, if data on criminals are entered into the database and video analytics methods will detect sweat There is an official criminal at the entrance to the store.

4. Organization of face control at large mortgages. An effective recognition system for detecting alarming information at remote security devices will help reduce or completely eliminate the occurrence of unnecessary events in a large number of people.

5. Organization of sales and targeted advertising. Based on the recognition and identification of a person, it can be determined how to display advertising that will potentially appeal to the client. At the same time, based on the collected information, it is possible to write off cash from the client’s account at the time of making another purchase without the participation of a person.

Regardless of the video stream processing algorithm, the software function for recognizing and identifying individuals follows the principle of aligning the scanned image with the standards in the database. When scanning is done on the go, it is enough to turn your face to the scanner while you are moving.

Essentially, recognition systems are computer programs that analyze images of people’s personalities to identify them. The program takes into account the image and displays such characteristics as the face of the eyes, the face of the nose, the edges of the cracks, on the basis of which a unique file is created, which is called a “template”. Based on the templates, the program compares the images with other images in the database, and then evaluates how similar the images are to each other. The primary means are to display signals from video cameras when identifying by appearance, or before removing photographs, which are stored in the water database.

This approach extends the visibility of the songs to the image captured from the video stream. A good indicator of the effectiveness of this speed code is that the system can effectively identify human identities from a distance of at least 10 meters from the video camera. In this case, recognition can be successfully implemented due to changes in basic physical parameters: change in combing, appearance of a beard, etc. During the period of time, recognition and identification does not have to overestimate the significance, for example, in the case of video monitoring of entrance doors and turnstiles. Another advantage is the video security characteristics. A number of types of IP cameras that may have the necessary characteristics are selected from each order.

In connection with the fact that the above descriptions can seriously interfere with the process of identification and recognition, the level of destructive recognition and identification is high. The problem of such a high value of the indicator of hypnotic recognition is also connected with the fact that between the fingers and the iris, our personalities change over time. Recognition systems are easily modified through a change in hairstyle, growth in the appearance or body, through the use by humans of the simplest methods of changing appearance, and also through the legacy of antiquity.

As a result of an investigation conducted by the National Institute of Standards and Technology (NIST), the results of which revealed that there was a level of fraudulent identification of subjects whose photographs were collected over 18 months In addition, it is higher than 43%. In this case, photography, vikorstan in the investigation, was raised in ideal minds, which is very important, some recognition programs are really bad at coping with the assessment of changes in brightness or where the cameras are and.

The image of an individual recorded in the template is subject to a number of factors, which indicate the strength of the identification and recognition systems. In the first place, the whole mind is illuminated and different occlusions, such as eyepieces or a mask, as well as the direction of rotation, the strength and the vision. Baby 1 shows the position of the head, which must be taken into account during the process of recognition and identification of the individual.

Malyunok 1 - Kutovy head position

The functioning of the person recognition and identification system from video recording begins with the development of people's patterns. As a rule, templates consist of two-dimensional images or frames extracted from a video stream. Below is a schematic representation of the process of registering a new template for the recognition and identification system.

Figure 2 - Schematic representation of the process of registering a new template in the recognition and identification system

Nowadays, tridimensional models are increasingly being used, which perform triangulation when sampling two-dimensional images or tridimensional images, when the RGB array is supplied with a depth matrix D, supported by an infrared sensor. Three-dimensional models can also be removed using the triangulation method, which will be combined with photographs taken using special equipment.

A way to force the pattern to lie in the methods used in the recognition and identification system, the format of the output data, or the task of the operating system. The most important stages that describe the formation of the pattern of the recognition and identification system of the characteristics presented to the baby 3

Figure 3 – The process of forming a template on the application of the recognition and identification system.

1. At the stage of processing, the detection of features and transformation of the visual areas to a different form takes place: wrapping (virdification), scaling, reorganization of channels, etc.

2. Another stage can include both searching for key points and submitting the pixel matrix of an individual in the character space. Under the signs, there are additional image descriptors, extracting the results of processing the output data.

3. At the last stage, the template is coded and entered into the model database. A database of templates has been formed and is then used as a set of standards.

Let us present a schematic representation of the principle of identification and recognition of individuals on baby 4, in which the following values ​​are shown:

1. Transferring the system image from a video stream.

2. Identification of the expression of the patterns evident in the base.

3.
Verification of the closest sign in the list and submitted to the system input.

Malyunok 4 - Schematic illustration of the principle of identification and recognition of individuals

Once the identification process is successful, the system returns the positive result that appears to the type of person in the displayed video stream to the identified subject.

Thus, systems for recognizing and identifying individuals from real-time video capture have a wide range of applications and function in a small way that is superimposed on the images they contain. As part of further investigation, we will analyze the main solutions for galusa recognition and identification of individuals from real-time video recording.

Today's area of ​​recognition and identification systems, video recording, covers practically the skin sphere of human activity. One of the key areas of detention is security activity, and the main objects of recognition are individuals and license plates of transport vehicles. Today there is a trend of increasing demand for complex functions - roc detection and extraneous objects, roc trajectory detection, rich chamber tracking, classification and identification objects, recognition of the situation, analysis of people’s behavior, etc.

The fight against terrorism and crime can be considered a critically important area in the development of recognition and identification systems based on video recording. In cases where images of individual criminals are stored in a database, and in places where there is a mass gathering of people (for example, airports, train stations, shopping malls, sports facilities), surveys are carried out in real time, the search for such systems is an effective method for identifying , who are in the middle of trouble .

Thus, the problems of this investigation lie in the established low unsolved problems of the stability of recognition algorithms until the influx of new minds and computing power, due to the rapid development of the galusia of stagnation of recognition systems and identification of video recordings in real time. It is necessary to develop a current design of an effective video recording system to recognize and identify events in real time, in order to provide clear guidance for the successful implementation of the advanced technology ї.

Based on the relevance of the identified problems, it is possible to draw conclusions about the extent of developing the project of an effective video recording system by recognizing and identifying features in real time mode.

Bibliographic mailing

Yurko I.V., Aldobaeva V.M. AREAS OF RESEARCH AND PRINCIPLES OF ROBOT SYSTEMS AND IDENTIFICATION OF REAL VIDEO FIXATION // International Student Scientific Newsletter. - 2018. - No. 2.;
URL: http://eduherald.ru/ru/article/view?id=18416 (date published: 01/01/2020). We would like to present to you the magazines that are available at the Academy of Natural Sciences

Biometrics is a set of methods and devices for identifying people that are based on their unique physiological and behavioral characteristics.

This type of identification can be used to prevent secure access to telephones, computers, ATMs, mobile phones, and so on.

Biometric power:

• finger bruises;
• geometry of the face;
• rainbow cover of the eyes;
• baby netkivka;
• voice;
• handwriting;
• friend on the keyboard;
• veins on the arms and hands.

Science 2.0 Identification of specialness

Advantages of biometric identification

Biometric security has a greater effect when used with passwords, smart cards, PIN codes, tokens or private key infrastructure technologies. This is explained by the ability of biometrics to identify not devices, but people.

Other security methods can lead to the loss or theft of information, which is open to illegal profiteers. An innovative biometric identifier, such as a fingerprint, is a key that prevents waste.

Classification of biometric methods

Depending on the type of information collected, biometric identification is divided into:

• Static methods, based on unique powers, given to people from among the people and unknown to others. Physiological indicators (the geometry of the palm or the papillary finger of the fingers) are unchanged for humans.”
• Dynamic methods based on behavioral (or dynamic) characteristics of particularity. These are the peculiarities of hidden hidden hands when performing any actions (prompts, signatures, keyboard typing dynamics). Such behavioral indicators sense the influx of cerebral and even cerebral mental officials. Through their variability, biometric symbols are responsible for their changes.

Methods for identifying characteristics behind biometric parameters

This recognition method is the most extensive. What is unique is the uniqueness of the papillary visceral fingers of a skin person. A special scanner captures images of the fingerprint. It is transformed into a digital code and is set to the template entered earlier.

The identification process takes no more than a few seconds. The smallest part of the development of this method lies among advanced people who are not afraid to give up information about their fingertips. The counterargument of the retail equipment lies in the fact that information about the fingerprint is not saved, but only a short identification code is saved, requiring the finger to be entered and does not allow the user to leveling. The advantage of this method is its simplicity, reliability and ease of use.

Based on the shape of the hand

This static method is based on the actual shape of the hand. It is also a unique biometric parameter of a person. A special device allows you to remove the trivial look of the pencil. As a result, it is necessary to create a unique digital code that identifies a person.

This method, in terms of its technology and accuracy, can be compared with the method of pressing the finger, although the device itself for implementing the method takes a lot of place. It is extremely unlikely that there are two identical hands that move according to the same geometry, although the hands change with each other.

Today, identification based on the geometry of the hand is stagnated by legislators, hospitals, international airports, etc.

Iris authentication

The basis of this method is the viscosity of the eye on the iris of the eye. It requires a camera to capture images in sufficient detail, and a special software program to view the captured image on the outside. Accordingly, a digital code is created that serves to identify a person.

The advantage of scanners is that people do not have to focus on the target, the fragments of the iris plasma are concentrated on the surface of the eye. Scanning is possible at a distance of less than 1 m. This is ideal for scanning, for example, at ATMs.

Eye mesh identification

The grids are scanned using low-intensity infrared light directly to the blood vessels of the posterior wall of the eye through the sine. Mesh scanners are widely used in access systems for sensitive objects, so they may not allow incorrect access permissions. Corrections can be explained by movements of the head away from the standard position and incorrect focusing of the eyes on the eyes of the light.

The twins' capillary network is becoming increasingly severe. Therefore, this method can be successfully used to identify specificity.

In a few such systems, a psychological factor can be introduced: not every person can look into a dark hole where one can shine into an eye. In addition, these systems are sensitive to incorrect orientation of the reticle, so you need to carefully follow the eye positions before opening.

Form individuals as an object for identification

This static method of identification applies to a created two or threefold image of a human individual. Using a camera and specialized software, the contours of eyes, lips, eyebrows, nose, etc. are added to the imaged face. Then calculate the differences between these elements and other parameters. Behind these records, an image is created that is transformed into digital form.

This method is being developed most dynamically in the biometrics industry. Its advantage lies in the fact that it does not require special expensive equipment. A personal computer and video camera are sufficient. In addition, daily physical contact with devices. There is no need to rush into anything or hesitate, especially paying attention to the system’s operation.

Recognizing handwriting

The basis for identifying handwriting is the uniqueness and stability of the person's skin. The characteristics are simulated, translated into digital form and subject to computer processing. So for leveling, it is not the leaf as a product that is chosen, but the process.

Two methods of data processing have been expanded: primary alignment and dynamic verification. The first one is unreliable, because the signature will not be the same again. This method can lead to a great number of benefits. Dynamic verification applies to folding calculations. With this method, the parameters of the signing process itself are registered in real time: the smoothness of the hand on different parts, the strength of the pressure and the difficulty of the various stages of signing. This includes a detail, it is impossible to accurately copy the author’s handwriting to the signature.

Recognition behind the keyboard handwriting

This method is similar to the one described above, except that the signature is replaced with a code word, so that a simple keyboard is not necessary for installation. The main identifying characteristic is the dynamics of typing the keyboard code word.

Based on current research, the keyboard handwriting has a distinct stability, which is why one can clearly identify its peculiarity. The output data is an hour between the pressure of the keys and their relaxation. Moreover, the hour between the pressures shows the tempo of the work, and the morning shows the style of the work, a smooth pressure and a sharp blow.

Initially, at the filtering stage, data about “service” keys is visible – functional, cursor control, etc.

Then you can see the following characteristics of the koristuvach:

• number of grants in the recruitment process;
• an hour between pressing keys;
• Smoothness of the set.
• an hour to adjust the buttons;
• Arrhythmia when typing .

Understanding the voice

The biometric method of voice identification is manual. The reasons for this stagnation are the widespread expansion of telephone networks and the introduction of microphones in computers. You can little by little take into account factors that affect recognition: interruptions in microphones, noise that can be heard, confusion in the process of recognition, different emotional states of people during identification, etc.

The main thing in everyday voice authentication devices is to select parameters that best describe the individuality of the voice. These parameters of the signal are called signs of individuality. Such signs, in addition to data about the peculiarity of the voice, are to blame for the mothers of other authorities. For example, the stench is responsible for easily fading and little exposure to noise and noise. In addition, they are responsible for the stability of the hour and the operation of imitation.

The system has been developed based on the method of combined analysis of voice and facial expressions. It turns out that the facial expression of the one who speaks teases only him and the other person will hear the same words of another person.

Thermographic monitoring of facial arteries and veins

Identification of a person by person will say goodbye in order to move into the infrared range of light colors. Thermography of an individual that is identified reveals the uniqueness of the development of arteries that provide the skin with blood. There is no power supply for these biometric devices, and they don’t need any light to absorb the stink due to temperature changes. The effectiveness of recognition depends on overheating and hypothermia of an individual, natural characteristics, plastic surgeries, and debris from changing the internal structure of vessels.

The method of facial thermography can separate twins, the facial blood vessels of which vary greatly.

A long-range infrared camera is specialized for this method of identification.

Identification behind the veins of the arm

There are devices on the biometric market that were used to analyze the individual distribution of veins in the arms. Take with respect the tiny veins that are spread out on the back side of the hand clenched into a fist. A television camera with infrared illumination monitors the small veins. When the image is entered, a binarization is carried out that shows the veins. A similar product is produced by the English company Vinchek.

Prospects for biometrics

The dominant method of identifying an individual is the recognition of fingertips. There are two main reasons for this:

• In many countries, the transition to passports with biometric data has begun;
• development of updated models of fingerprint scanners for use in small devices (phones, desktop PCs, laptops).

This expansion can be found in the signature identification sector in connection with the wide range of digital electronic signatures. Voice recognition can also gain momentum in the implementation of great projects among everyday intellectuals.

The main forecasts are based on the fact that the deployment of biometric security devices will result in an avalanche outbreak. The fight against global terrorism is a matter of practical vigor in any way in this sphere. Finally, the intensive development of multimedia and digital technologies and their reduction in cost will allow the development and introduction of fundamentally new identification systems.

All biometric technologies are currently undergoing development and some of them are considered promising:

1. thermogram of an individual in the infrared range;
2. DNA characteristics;
3. spectroscopy of the skin of the fingers;
4. vіdbitki dolon;
5. ear shape;
6. parametri hodi lyudini;
7. individual odors;
8. the amount of saltiness of the skin.

These methods of biometric identification can be developed today. It is possible that soon the smell will pass from scientific research to commercial technologies.

The rest of the time on Habr there is a lack of articles dedicated to Google’s personal identification systems. To be honest, the rich people are so ignorant of journalism and seemingly incompetent. And I wanted to write an article about biometrics for Garn, but it’s not my first choice! A couple of nasty articles about biometrics on Habriya - but they won’t be long and short. Here I will try to briefly describe the hidden principles of biometric identification and the current achievements of mankind in its nutrition. This includes the identification of individuals.

The article is a continuation, as, in essence, it is a prequel.

As a basis for the article, it will be combined with other publications in the magazine (BDI, 2009), revised to suit current realities. Colleagues are still absent from Habré, but he supports the publication of the reworked article here. At the time of publication, this article was taking a short look at the current market for biometric technologies, which we were doing before launching our product. Estimates about stagnation, hanging in another part of the article, are based on the thoughts of people who have stolen and sold products, as well as on the thoughts of people who are engaged in the development of biometric systems in Russia and Europe.

Secret information

Let's finish with the basics. In 95% of cases, biometrics are based on mathematical statistics. And matstat is an exact science, the algorithms behind which are tested through: both in radars and in Bayesian systems. As two main characteristics of any biometric system, one can take advantage of the first and the other kind). Theoretically, radars are called “hibna alarm” and “target overshoot”, and biometrics are most commonly known as FAR (False Acceptance Rate) and FRR (False Rejection Rate). The first number characterizes the possibility of an equal comparison of the biometric characteristics of two people. The other is confidence in the access of the person who has access. The system is shorter, lower FRR values ​​at the same FAR values. In some cases, the EER characteristic is determined and equalized, which means the point at which the FRR and FAR graphs interchange. Ale vona is far from being representative. You can marvel at the report, for example.
You can mean this: If the system characteristics do not provide FAR and FRR for open biometric databases, then the manufacturers do not declare their characteristics, and this system is most likely unsuitable or much weaker than its competitors.
Not only FAR and FRR indicate the strength of the biometric system. If it were only so, then the wired technology would be the recognition of people's DNA, for which FAR and FRR would be reduced to zero. It is obvious that this technology is not stagnant at the current stage of human development! We have collected a number of empirical indicators that allow us to evaluate the strength of the system. “Durable resistance” is an empirical characteristic that indicates how easy it is to fool a biometric identifier. “Durability to extreme consistency” is a characteristic that empirically evaluates the durability of a system’s operation based on various external factors, such as changes in lightening or ambient temperature. “Simplicity of Vikoristan” shows how difficult it is to use a biometric scanner and how identification is possible “on the go.” An important characteristic is “Robot Speed” and “System Variety”. It is also important to remember that the biometric characteristic of a person can change over time, which is why it is unstable - this is a negative point.
A large number of biometric methods are controversial. The main methods used to determine the static biometric characteristics of a person are identification by the papillary veins on the fingers, the iris, facial geometry, reticulum of the eye, small veins of the hand, hand geometry. There is also a family of methods that detect dynamic characteristics: identification of voice, handwriting dynamics, heart rate, walking. Below is a breakdown of the biometric market, which is part of the reason for this. For every other person, the data varies by 15-20 hundred, so there is no estimation. So here, under the concept of “hand geometry,” there are two different methods, which will be discussed below.

The statistics have the same characteristics as those found in access control systems (ACS) or similar departments. When I look at my advantage, I look at the most static characteristics. Of the dynamic characteristics, the recognition of the voice may have some statistical significance (equal to the larger static algorithms FAR~0.1%, FRR~6%), but in ideal minds.
In order to understand the consistency of FAR and FRR, you can evaluate how often milk runs are produced by installing an identification system at a passing organization with a number of personnel N in particular. The consistency of the fingerprint removed by the scanner for the N data base is the same as FAR∙N. Today, you must pass through the access control point very close to N in particular. The same amount of compensation per working day is FAR∙(N∙N). Of course, for the purposes of the identification system, the consistency of a meal in one hour can change greatly, but if we accept one meal throughout the working day, then:
(1)
It is also clear that the stable operation of the identification system for FAR=0.1% =0.001 is possible for a staff of N≈30.

Biometric scanners

Today, the concepts of “biometric algorithm” and “biometric scanner” are not necessarily interrelated. The company can release these elements one by one, or at once. The greatest differentiation of scanner vendors and software vendors is achieved in the fingerprint biometrics market. The newest 3D scanner on the market. In fact, the pressure of differentiation largely reflects the dissolution and saturation of the market. The more choice there is, the more the topic is elaborated and brought to perfection. Different scanners have a different set of capabilities. The main purpose is a set of tests to verify the details of the biometrics object. For finger scanners, you can check the relief and temperature, for eye scanners you can check the accommodation zone, for face scanners you can check the face.
Scanners are already putting a lot of emphasis on FAR and FRR statistics. In some situations, these numbers can change tenfold, especially in real minds. Therefore, the characteristics of the algorithm are given for some “ideal” basis, or simply for a well-suited one, where unsharp and distorted frames are thrown out. There are not enough algorithms to honestly specify both the base and the FAR/FRR data behind it.

And now more details about skin technology

Fingertips

Fingerprinting (finger recognition) is the most widely used biometric method for identifying characteristics today. The catalyst for the development of the method was the widespread development of criminology in the 20th century.
Human skin has a unique papillary pattern of the fingertips, which is why identification is possible. Set the algorithms to detect characteristic points on the fingertips: the end of the line of the finger, the straightening of the line, single points. Additionally, information is obtained about the morphological structure of the finger: the clear position of the closed lines of the papillary vein, “arched” and spiral lines. The peculiarities of the papillary image are converted into a unique code, which preserves the information content of the image. And the “finger codes” are stored in the database, which is used for searching and updating. The time it takes to convert a finger image into a code and identification does not exceed 1s, depending on the size of the base. An hour spent on raising your hands is not covered by insurance.

In addition to the FAR and FRR data, statistical data from the VeriFinger SDK was collected using the DP U.are.U finger scanner. Over the last 5-10 years, finger recognition characteristics have not progressed much forward, so pointing the numbers is not good enough to show the average values ​​of current algorithms. The VeriFinger algorithm itself has been a part of the International Fingerprint Verification Competition, where finger recognition algorithms have been developed.

The typical FAR value for the finger finger recognition method is 0.001%.
From formula (1) it is assumed that stable operation of the identification system FAR=0.001% is possible for a number of personnel N≈300.

Advantages of the method. High reliability - statistical indicators of the method of redundancy for indicators of methods of identification by person, voice, signature. Low versatility of devices to scan fingerprint images. The scanning procedure is simple.
Shortcomings: the papillary ridge of the finger is easily damaged by small marks and cuts. People who have used scanners in enterprises with a staff of close to hundreds of people claim a high level of scanner technology. Many scanners are inadequately installed until the skin is dry and do not allow older people to pass through. At the hour of deployment at the remaining MIPS installation, the head of the safety service of a large chemical enterprise confirmed that their attempt to introduce finger scanners at the enterprise (scanners from various systems were tried) failed - minimal action chemical reagents on the fingers of spivrobitniks called out the failure of the security systems of the scanners - the scanners stunned their fingers. There is also a lack of security in the details of the image format, which is often used by a wide-ranging method. Of course, not all scanners can be fooled using methods from the ruins of Legends, but still. For some people with “unusual” fingers (specific body temperature, vulgarity), the confidentiality of access can reach 100%. The number of such people varies from hundreds of hundreds for expensive scanners to ten hundreds for inexpensive ones.
Of course, it means that a large number of shortcomings are affected by the wide range of the system, but these little ones move around and stinks appear even more often.

Market situation

Currently, finger recognition systems account for half of the biometric market. A number of Russian and foreign companies are engaged in the development of access control systems based on the fingerprint identification method. Through those that are directly one of the oldest, they have gained the greatest expansion and are today the most developed. The scanners of the fingertips went through a long process before being repainted. Modern systems are equipped with various sensors (temperature, pressure, etc.) to improve the protection of particles. Every day the systems become more simple and compact. In fact, the researchers have already reached the singing boundary in this field, and there is no way to develop the method further. In addition, most companies create ready-made systems that are equipped with everything necessary, including security software. In this case, integrators simply do not need to select a system on their own, since it would not take more time and effort than to buy a ready-made and inexpensive system, since the choice will be effectively wide.
Among the foreign companies that deal with finger recognition systems, one can include SecuGen (USB scanners for PCs, scanners that can be installed in enterprises or built into locks, SDK and software for linking the system with a computer); Bayometric Inc. (fingerprint scanners, TAA/Access control systems, fingerprint SDKs, embedded fingerprint modules); DigitalPersona, Inc. (USB scanners, SDK). In Russia, this industry operates the following companies: BioLink (fingerprint scanners, biometric access control devices, software); Sondu (fingerprint scanners, biometric access devices, SDK); SmartLock (fingerprint scanners and modules) etc.

Raiduzhna obolonka

The iridescent membrane of the eye is a unique characteristic of a person. The baby of the iris is formed in the eighth month of intrauterine development, remains stabilized in about two years and practically does not change throughout life, except after severe injuries or severe pathologies ій. The method is one of the most accurate among biometric methods.
The system for identifying features based on the iris is logically divided into two parts: the storage device for the image, its primary processing and transmission, and the calculation of the subsequent alignment of the image with the images in the database. , which transmits the command for the admission of the commander to the annex.
The hour of initial image processing in current systems is approximately 300-500 ms, the speed of alignment of the image received from the base ranges from 50000-150000 adjustments per second on a standard PC. Such smoothness of leveling does not interfere with the stagnation of the method in large organizations during the time of disruption in access systems. With the use of specialized calculation processors and optimization algorithms, it becomes possible to identify a person among the inhabitants of an entire region.
I can tell you that I am very proactive and positive about this method, since we launched our startup in this field. A small paragraph will be dedicated to a small self-promotion.

Statistical characteristics of the method

The characteristics of FAR and FRR for the iris of the eye are the best among the class of current biometric systems (except, perhaps, by the method of recognition by the retina of the eye). The article shows the characteristics of the iris recognition library for our algorithm - EyeR SDK, which is similar to the VeriEye algorithm verified from these very databases. The basics of the CASIA company were analyzed and captured by a scanner.

The typical FAR value is 0.00001%.
This is consistent with formula (1) N≈3000 - the number of personnel of the organization, if the identification of the employee is expected to be stable.
Here it is important to note the important feature that distinguishes the iris recognition system from other systems. With a separate 1.3MP camera, you can capture two eyes in one frame. Since the values ​​of FAR and FRR are statistically independent values, when recognized by two eyes, the FAR values ​​are approximately equal to the square of the FAR value for one eye. For example, for a FAR of 0.001% with a total of two eyes, the consistency of the cut tolerance is more than 10-8%, with an FRR of every two eyes, a lower FRR value for one eye with FAR=0.001%.

Advantages and disadvantages of the method

Advantages of the method. The reliability of the algorithm is statistical. Buried images of the iris can be captured from a distance of several centimeters to several meters, but the person has no physical contact with the device. The iridescent membrane is protected from damage - it will no longer change in time. It is also possible to use a high number of methods to protect the particles.
Some problems with the method. The price of a system based on the iris is higher than the price of a system based on a recognized finger or a recognized individual. Low availability of ready-made solutions. Any integrator who comes to the Russian market today and says “give me a ready-made system” will probably end up in trouble. More expensive turnkey systems are sold here, installed by great companies such as Iridian or LG.

Market situation

At the moment, the technology of identification by the iris of the eye on the light biometric market is being used for various purposes from 6 to 9 hundredths (at the same time as the technology of recognition of fingertips and borrow over half the market). It should be noted that from the beginning of the development of this method, its importance over the market was enhanced by the high level of installed components necessary to select an identification system. With the development of digital technologies, the integrity of the surrounding system began to decrease.
The leader in software development in this country is Iridian Technologies.
The entry of a large number of printers into the market is limited by the technical complexity of scanners and, as a result, their high productivity, as well as the high price of software through Iridian’s monopoly on the market. These factors have allowed only great companies to develop in the field of iris recognition, which are mostly already engaged in the development of various components associated with the identification system (high-resolution optics, miniature cameras with infrared switches too). Examples of such companies could be LG Electronics, Panasonic, OKI. They entered into an agreement with Iridian Technologies, and as a result of their extensive work, the following identification systems appeared: Iris Access 2200, BM-ET500, IRISPass OKI. Further attention was paid to the detailed model of systems due to the technical capabilities of these companies to independently develop in this area. It should be said that the reinsurance companies also developed the software accordingly, and as a result, the finished system gives priority to the software of Iridian Technologies.
On the Russian market, the products of foreign companies are “outweighed”. If you want, you can buy it, it’s important. During this troubling hour, the Papillon firm sang to everyone whose stench could be identified across the rainbow. However, representatives of RosAtom - their middle-of-the-road buyer, for whom the system was built, say that this does not indicate effectiveness. It seems that there was also a Russian company that built iris scanners. I can’t remember Nina’s name anymore. The stink algorithm was purchased from someone, perhaps from the same VeriEye. The scanner itself is a system 10-15 years old, and is by no means contactless.
In the rest of the world, a number of new detectors have entered the world market in connection with the completion of the first patent for recognizing people by sight. The greatest trust among them, in my opinion, belongs to AOptix. Accepted their previous documentation does not raise suspicion. Another company is SRI International. Apparently, at first glance, people who have worked on iris recognition systems, their videos look even lame. Although I’m not surprised because the stench actually sinks in. And this system does not show data on FAR and FRR, and, most importantly, it is not protected from data.

Recognition by appearance

There are no methods for recognizing the geometry of the surface. All this is based on the fact that the shape of the skull of a person’s skin varies from person to person. This area of ​​biometrics seems to be very interesting, because we recognize one another in the first place by appearance. This area is divided into two directions: 2-D recognition and 3-D recognition. Each of them has advantages and shortcomings, but another type of sphere of stagnation can be stored in large quantities, presented to a specific algorithm.
Let's briefly talk about 2-d and move on to one of the most popular methods today - 3-d.

2-D identification recognition

2-D facial recognition is one of the most statistically ineffective biometric methods. Having appeared long ago and having stagnated mainly in criminology, which fueled his development. Then came the computer interpretation method, the result of which became the most reliable, but, insanely, compromising with skin fate, it is more compromising with other biometric methods of identifying features. Currently, through poor statistical indicators, we are stuck in multimodal or, as they call it, cross-sectional biometrics, or in social measures.

Statistical characteristics of the method

For FAR and FRR, wiki data for VeriLook algorithms. Again, for current algorithms you can use simple characteristics. Sometimes you can try algorithms with an FRR of 0.1% with a similar FAR, but the basis behind some stench is removed even dubiously (the background is sharp, however, it is revealing, however, the cleaning, clarification).

The typical FAR value is 0.1%.
From formula (1) we deduce N 30 - the number of personnel of the organization, in which case the identification of the professional worker is expected to be stable.
Apparently, the statistical evidence for the method is modest: it does not detract from the advantage of the method that it is possible to capture individuals in populated areas. It’s funny to think how many times in the world the project is financed by identifying criminals through video cameras installed in public places. Over the past dozen years, the statistical characteristics of the algorithm have not been improved, and the number of such projects has grown. I would like to say that the algorithm is completely suitable for keeping people in the crowd through faceless cameras.

Advantages and disadvantages of the method

Advantages of the method. With 2-D recognition, most biometric methods are used, no installation is required. Reliable recognition is possible at significant locations in front of the camera.
The underdogs. Low statistical reliability. It is possible to hang out before lightening (for example, it is not possible to register individuals who enter the streets of people on a sleepy day). For many algorithms, any external changes are unacceptable, such as, for example, eyepieces, beards, and other trimming elements. The obov'yazkovo front image is oblique, with very small embellishments. Many algorithms do not guarantee the possibility of changing facial expressions, so that the appearance may be neutral.

3-D identification recognition

The implementation of this method is complicated. Regardless of this, at this time there is a use of non-invasive methods with 3-D recognition of evidence. It is impossible to compare the methods one by one, the fragments of the stench vikorist from different scanners and databases. Not all of them show FAR and FRR; there are completely different approaches.
The transition from 2-d to 3-d method is a method that implements accumulated information about a person. This method has better characteristics than the 2D method, but also because it only requires one camera. When a subject is added to the database, the subject turns his head and the algorithm merges the images together, creating a 3D template. And when the victor is recognized, a small number of frames are added to the video stream. This method will soon be brought to experimental and implementation for ACS systems, I have never done before.
The most classic method is the design template method. This is because a mesh is designed around the object. Then the camera takes pictures at a speed of tens of frames per second, and the images are captured using a special program. Anything that falls on a curved surface bends - the greater the curvature of the surface, the more it bends. At the same time, a source of visible light was stagnated, which is supplied through the blinds. Then the visible light was replaced by infrared, which has a low value. Therefore, at the first stage of processing, images are displayed in which the individual is not visible or there are foreign objects that are important for identification. Behind the removed photographs, a 3-D model of the individual is updated, in which unnecessary defects (combing, beard, hair and eyepieces) are visible and visible. Then the model is analyzed - anthropometric features are seen, which as a result are recorded in a unique code that is entered into the database. The hour of storage and image processing is set to 1-2 seconds for the largest models.
This is how the method of 3-D recognition of images captured from several cameras is gaining popularity. The Vocord company can use this application with its 3D scanner. This method provides positioning accuracy, similar to the patterns of the distributors, rather than the method of designing the template. Until I figure out FAR and FRR, I’d like to use their power base - I won’t believe it! Ale yogo are destroying already 3 rocks, but the destruction is not yet visible on exhibitions.

Statistical indicators of the method

There are no new data about FRR and FAR for algorithms of this class on the developers’ websites. For the shortest models from Bioscript (3D EnrolCam, 3D FastPass), use the template design method with FAR = 0.0047%, FRR set to 0.103%.
It is important that the statistical reliability of the method can be compared with the reliability of the fingerprint identification method.

Advantages and disadvantages of the method

Advantages of the method. There is no need to contact the scanning device. Low sensitivity to current officials, both in the person himself (appearance of eyepieces, beard, change in hairstyle), and in his features (lightening, turning of the head). High level of reliability, which can be equaled by fingerprint identification.
Some problems with the method. Dear owner. The complexes that were currently on sale were converted for the price of iris scanners. Change facial expressions and switch to accusations to improve the statistical reliability of the method. The method is still insufficient for good analysis, especially in comparison with fingerprinting, which has long been stagnant, which complicates its wider stagnation.

Market situation

Recognition of geometry reveals up to “three great biometrics” simultaneously from recognition of fingertips and iris. It is necessary to say that this method can be expanded to give it an advantage over recognition by the rainbow eye. Therefore, the power of technology to recognize the geometry of an individual in the global market of the light biometric market can be assessed in the range of 13-18 hundredths. Russia also shows greater interest in this technology, for example, to identification by iris. As was predicted earlier, there are no human 3-D recognition algorithms. Most companies will develop ready-made systems that include scanners, servers and software. However, there are those who use the same SDK. Today, you can identify the following companies that are engaged in the development of this technology: Geometrix, Inc. (3D facial scanners, PZ), Genex Technologies (3D facial scanners, PZ) in the USA, Cognitec Systems GmbH (SDK, special computing, 2D cameras) in Germany, Bioscrypt (3D facial scanners, PZ) – a subsidiary of the American company L- 1 Identity Solutions.
In Russia, the company directly operates Artec Group (3D scanners and PZ) - a company whose head office is located in California, and development and production are carried out in Moscow. Also, a number of Russian companies are using 2D individual recognition technology - Vocord, ITV, etc.
In the field of recognition of 2D individuals, the main subject of development is the security program, because Secondary cameras cope well with the accumulation of revealing images. The current state of image recognition has apparently gone into oblivion - for many years there has been virtually no increase in the statistical indicators of algorithms. This galusa is engaged in a systematic work on milking.
3D detection of infectious diseases is a hugely exciting area for researchers. There are no teams and regularly hear about new discoveries. Bezlіch works at the station “axis-axis and let’s go.” However, there are still no old propositions on the market for the rest of the time, the choice has not changed.
One of the important points that I sometimes ponder and which, perhaps, is Habr's answer: is it possible to improve the accuracy of kinect to create such a system? Projects from drawing 3D models of people through it entirely.

Recognition behind the veins of the hand

The price of new technology in the field of biometrics began to become widespread around 5-10 years ago. The infrared camera takes pictures of the outer or inner side of the hand. Small veins are formed due to the fact that hemoglobin in the blood is lost due to infrared vibration. As a result, the intensity level changes and the veins appear as black lines on the camera. A special program creates a digital package based on the extracted data. There is no need to contact the person with the scanning device.
The technology can be compared for its reliability with the recognition of the iris of the eye, where we overturn, and where we compromise.
The FRR and FAR values ​​are calculated for the Palm Vein scanner. According to the retailer's data, with a FAR of 0.0008%, the FRR becomes 0.01%. The exact schedule for many values ​​is not visible to each company.

Advantages and disadvantages of the method

Advantages of the method. There is no need to contact the scanning device. High reliability - statistical indicators of the method can be compared with those of the iris. A characteristic has been captured: in addition to all the points made, this characteristic is very important to take away from a person “on the street”, for example, taking a photograph of him with a camera.
Some problems with the method. It is unacceptable for the scanner to be illuminated by halogen lamps or halogen lamps. Any age-old illness, such as arthritis, greatly increases FAR and FRR. The method of fewer changes is similar to other static biometric methods.

Market situation

Recognition of small hand veins is complemented by new technology, and the volume associated with this product in the light market is small and amounts to around 3%. However, there is increasing interest in this method. The truth is that, being quite accurate, this method does not require such expensive equipment as, for example, methods for recognizing the geometry of the face or the iris. There are a lot of companies currently conducting research in this area. So, for example, with the help of the English company TDSi, a software program was developed for the PalmVein biometric vein reader, presented by Fujitsu. The scanner itself was developed by Fujitsu in order to combat financial fraud in Japan.
Also in the field of identification of small veins are the following companies: Veid Pte. Ltd. (scanner, software), Hitachi VeinID (scanners)
I don’t know of any Russian companies that are involved in this technology.

Sitkivka oka

Until recently, it was believed that the most reliable method of biometric identification and identity authentication was based on a scanned eye reticle. You should place your best identification behind the iris and veins of the arm. The scanner reads the tiny capillaries on the surface of the eye retina. The mesh has a stable structure, unchanged over time, even as a result of illness, for example, cataracts.
The mesh scan is performed using low-intensity infrared light directly through the stenosis to the blood vessels on the posterior wall of the eye. Mesh scanners have become widely used in access control systems for particularly sensitive objects, since they have one of the fewest opportunities available to registered accountants and practically do not allow access permission.
Unfortunately, there are few difficulties associated with using this method of biometrics. The scanner here is a very complex optical system, and it takes a lot of time for people not to collapse until the system is aimed, which screams in an unpleasant way.
According to data from the EyeDentify company for the ICAM2001 scanner, with FAR=0.001% the FRR value becomes 0.4%.

Advantages and disadvantages of the method

Advantages. High level of statistical reliability. Due to the low width of the systems, there is little opportunity for their method of “deception” to develop.
The underdogs. Foldable under the hour vykoristanny system with a high hour of cutting. The system's varity is high. The presence of a wide market position results from the lack of intensity in the development of the method.

Hand geometry

This method, to expand the scope of the 10th century, and leaving the field of criminology with the remaining forces is on the decline. It is based on the selected geometric characteristics of the hands: the length of the fingers, the width of the knees, etc. This method, like the reticle of the eye, is dying, and since the fragments have many lower characteristics, we will not introduce any further description.
It is also important that in vein recognition systems, geometric recognition methods are used. Ale, we have never been able to sell such a clearly stated product. Just as often, when veins are recognized, there are signs of the thigh, just as when recognized by geometry, there are signs of fingers.

A little self-promotion

In no time, we have dismantled the disgusting algorithm of recognition by sight. But at that time such a high-tech thing was not needed in this region, and we didn’t want to go to bourgeoistan (where we were asked to after the first statistics). Ale raptovo, through the second fate, investors were found who wanted to own a “biometric portal” - a system that had 2 eyes and vicorized the color of the warehouse iris shell (for which the investor would have a light patent). Vlasna, now we are busy. This is not an article about self-PR, but a short lyrical introduction. If anyone needs a little bit of information, in the future, when we enter the market (or we don’t), I’ll write a few words about the ups and downs of the biometric project in Russia.

Visnovki

In the class of static biometric systems there is a wide choice of systems. How can I vibrate them? Everything could be stored until the security system. The most statistically reliable and detailed access systems are access systems for the iris and arm veins. In the first place, there is a wider market for propositions. There is no limit. Biometric identification systems can achieve astronomical precision. The cheapest and simplest ones are from Vikoristan, and those that have good statistics, and admission systems on your fingers. Tolerance for 2D individuals is easy and cheap, but the area of ​​stagnation can be demarcated through poor statistical indicators.
Let's take a look at the characteristics of the skin's systems: resistance to chipping, resistance to dovkill, ease of wear, versatility, fluidity, stability of the biometric sign at an hour. We assign ratings from 1 to 10 for the skin graph. The closer the score is to 10, the better the system is. The principles for choosing assessments are described on the basis of statistics.


We will also look at the relationship between FAR and FRR for these systems. This relationship is determined by the efficiency of the system and its breadth.


Remember that for the iris shell it is possible to increase the accuracy of the system almost quadratically, without spending an hour, by folding the system into two eyes. For the fingerprint method - by combining several fingers, and identifying veins, by combining two hands, and this is also possible only with more hours, which is spent when working with a person.
Having summarized the results for the methods, we can say that for medium and large objects, as well as for objects with maximum security, use the iridescent membrane as a biometric access and, possibly, recognition along the veins of the arms. For facilities with a large number of personnel, up to hundreds of people, fingertip access will be optimal. Recognition systems based on 2D images are even more specific. They may be needed in cases of seizures, if recognition involves physical contact, otherwise it is impossible to install a system for monitoring the iris. For example, if it is necessary to identify a person without their participation, using a camera or an external detection camera, or perhaps only for a small number of subjects in the database and a small number of people, such as Remove the camera.

Notes for young technicians

Some vendors, such as Neurotechnology, have demo versions of the biometric methods they produce available on their website, so you can even connect them and play. For those who want to delve into the problem more seriously, I can please you with one book, which I read in Russian - “Handbook of Biometrics” by R.M. Ball, J.H. Connell, S. Pankanti. There are a lot of algorithms and mathematical models there. Not everything is perfect and does not indicate urgency, but the base is poor and voluminous.

P.S.

In this article I do not go into the problem of authentication, but rather about identification. In principle, due to the FAR/FRR characteristics and the possibility of details, you can ask yourself for all the necessary steps for authentication.