Biometricheskaya identification in the integrated systems of safety
Traditional methods of identification of the person in which basis there are variou identification cards, keys or unique data, such as, for example, the password are not reliable in that degree which is required for today. Natural step to increase of reliability of identifiers of a steel of attempt of use of biometric technologies for safety systems.
The range of the problems, which decision can be found with use of new technologies, is extremely wide:
- To prevent penetration of malefactors on protected territories and in premises at the expense of a fake, thefts of documents, cards, passwords;
- To limit access to the information and to supply personal responsibility for its safety;
- To supply the admission to responsible objects only the certificated experts;
- To avoid an overhead charge connected with operation of monitoring systems of access (a card, keys);
- To exclude the inconveniences connected with uterey, damage or elementary zabyivaniem keys, cards, passwords;
- To organise the account of access and attendance of employees.
To working out of technologies for recognition of images under various biometric characteristics have started to attend already for a long time, the beginning was necessary in 60th years. Our compatriots have achieved significant successes in working out of theoretical bases of these technologies. However practical results are received basically in the west and only "yesterday". Capacity of modern computers and advanced algorithms have allowed to create products which under the characteristics and a parity became accessible and interesting to wide sections of users.
The idea to use individual characteristics of the person for its identification is not new. For today a number of technologies which can be involved in systems of safety for identification of the person on is known:
- To prints of fingers (both separate, and hands as a whole);
- To features (on the basis of optical and infra-red images);
- To iris of the eye of eyes;
- To voice;
- To other characteristics.
All biometric technologies have general approaches to the decision of a problem of identification though all methods differ convenience of application, accuracy of results.
Any biometric technology is applied stage by stage:
- Object scanning;
- Extraction of the individual information;
- Template formation;
- Comparison of a current template with a database.
The biometric system of recognition establishes conformity of concrete physiological or behavioural characteristics of the user to some set template. Usually biometric system consists of two modules: the module of registration and the identification module.
The registration module "trains" system to identify the concrete person. At a stage of registration the videocamera or other gauges scan the person to create digital submission of its shape. Scanning of the person lasts nearby 20 – 30 seconds therefore some images are formed. In an ideal case, these images will have slightly various foreshortenings and looks that will allow to obtain more exact data. The special program module processes this submission and defines prominent features of the person, then creates a template. There are some parts of the person which practically are not changed with the course of time, it, for example, the top outlines of eye-sockets, areas surrounding cheekbones, and mouth edges. The majority of the algorithms developed for biometric technologies, possible changes in a hairdress of the person as they do not use for the analysis of area of the person above border of growth of hair allow to consider. The template of the image of each user is stored in a database of biometric system.
The identification module receives the image of the person from a videocamera and will transform it to the same digital format in which the template is stored. The obtained data are compared with hranimyim in a database a template to define, whether there correspond these images one another. The degree of similarity required for check, represents a certain threshold which can be adjusted for various type of staff, capacities PC, time of days and of some other factors.
Identification can be executed in the form of verification, autentifikatsii or recognition. At verification identity of the received data and a template, hranimogo in a database is confirmed. Autentifikatsiya – confirms conformity of the image received from a videocamera to one of templates, stored in a database. At recognition if the received characteristics and one of hranimyih templates appear identical the system identifies the person with a corresponding template.
At use of biometric systems, especially recognition systems on the person, even at introduction of correct biometric characteristics not always the decision about autentifikatsii is true. It is connected with a number of features and, first of all, that many biometric characteristics can be changed. There is a certain degree of probability of an error of system. And at use of various technologies the error can essentially differ. For access monitoring systems at use of biometric technologies it is necessary to define, that it is more important not to pass "stranger" or to pass all “”.
The important factor for users of biometric technologies in safety systems is simplicity of use. The person which characteristics are scanned, should not test any inconveniences thus. In this plan the most interesting method is, unconditionally, technology of recognition on the person. However, in this case there are other problems connected first of all, with accuracy of work of system.
Despite obvious advantages, there is a number of negative biases against biometrii which often cause questions on, whether biometric data will be used for shadowing people and infringement of their right to a private life. Because of sensational applications and an unreasonable sensation the perception of biometric technologies sharply differs from a real state of affairs.
And still, use of biometric methods of identification has acquired a special urgency the last years. The especially sharply given problem was showed after events on September, 11th in the USA. The world community has realised degree of increase of threat of terrorism all over the world and complexity of the organisation of reliable protection by traditional methods. These tragical events have served as a starting point for attention strengthening to the modern integrated systems of safety. The opinion is well-known, that if the control at the airports was more religiously misfortunes could be avoided. And today search of other incidents guilty of a number could be essentially facilitated at use of modern systems of video observation in integration with systems of recognition of persons.
Methods of recognition of the person
Now there are four basic methods of recognition of the person:
- "eigenfaces";
- The analysis of "distinctive lines";
- The analysis on a basis of "neural networks";
- Method of "automatic processing of the image of the person".
All these methods differ with complexity of realisation and the application purpose.
"Eigenface" it is possible to transfer as "own person". This technology uses two-dimensional images in gradation grey which represent distinctive characteristics of the image of the person. A method "eigenface" are often used as a basis for other methods of recognition of the person.
Combining characteristics 100 – 120 "eigenface" it is possible to restore a considerable quantity of persons. At the moment of registration, "eigenface" each concrete person it is represented in the form of a number of factors. For a verification mode in which the image is used for check of identity, the "live" template is compared to already registered template, for the purpose of definition of factor of distinction. Distinction degree between templates also defines the identification fact. The technology "eigenface" is optimum at use in well shined premises when there is a possibility of scanning of the person en face.
Analysis technique of "distinctive lines" – most widely used technology of identification. This technology is similar to a technique "Eigenface", but in bolshey degrees is adapted for change of appearance or a mimicry of the person (smiling or frowning person). In technology of "distinctive lines” tens prominent features of various areas of the person, and with allowance for their relative site are used. The individual combination of these parametres defines features of each concrete person. The person of the person is unique, but is dynamical enough, since the person can smile, release a beard and moustaches, to put on points – all it increases complexity of procedure of identification. Thus, for example, at a smile some displacement of parts of the person possessed about a mouth is observed, that in turn will cause similar movement of adjacent parts. Considering such displacement, it is possible to identify unequivocally the person and at various mimic changes of the person. As this analysis considers local plots of the person, maximum deviations can be in limits to 25 ° in a horizontal plane, and approximately to 15 ° in a vertical plane and the powerful and expensive equipment requires enough, that accordingly reduces degree of distribution of the given method.
In a method based on a neural network, prominent features of both persons – registered and checked are compared on coincidence. "Neural networks" use the algorithm establishing conformity of unique parametres of the person of the checked person and parametres of a template, being in a database, the greatest possible number of parametres is thus applied. In process of comparison discrepancies between the person checked and a template from a database are defined, then the mechanism which by means of corresponding weight factors defines degree of conformity of the checked person to a template from a database is started. This method increases quality of identification of the person in difficult conditions.
Method of "automatic processing of the image of the person" – the most simple technology using distances and the relation of distances between easily defined points of the person, such as eyes, the nose end, mouth corners. Though the given method not so powerful as "eigenfaces" or "a neural network", it can be effectively enough used in the conditions of weak light exposure.
Recognition systems on the person, present in the market
For today a number of the commercial products intended for recognition of persons is developed. The algorithms used in these products, are various and it is for the present difficult to state an estimation what of technologies has advantages. Leaders at the moment are following systems: Visionic, Viisage and Miros.
- At the heart of appendix FaceIt of company Visionic the analysis algorithm of local signs developed at University of Rockefeller lays. One commercial company to the Great Britain integrated FaceIt into television anticriminal system under name Mandrake. This system searches for criminals on video data which act from 144 chambers united in the closed network. When identity is established, the system informs on it to the officer of safety. In Russia the representative of company Visionic is company "DanKom".
- One more leader in this area, company Viisage, uses the algorithm developed in the Massachusetsky institute of technology. The commercial companies and the state structures in many American states and in a number of other countries use company Viisage system together with identification certificates, for example, a driver’s licence.
- ZN Vision Technologies AG (Germany) offers a number of products in which the technology of recognition of persons is applied in the market. These systems are represented in the Russian market by company "Soling".
- In system of recognition of persons TrueFace of company Miros the technology of neural networks is used, and the system is applied in a complex of issue of cash of corporation Mr. Payroll also it is established in a casino and other pleasure establishments of many states of the USA.
In the USA independent experts had been held comparative testing of various technologies of recognition of persons. Results of testing are presented more low.
In practice, at use of systems of recognition of persons as a part of standard electronic security systems, the person who should be identified is supposed, that, looks directly in the chamber. Thus, the system works with rather simple two-dimensional image, that considerably simplifies algorithms and reduces intensity of calculations. But even in this case the recognition problem nevertheless is not trivial, as algorithms should consider possibility of change of level of illumination, look change, availability or absence of a make-up or points.
Reliability of work of system of recognition of persons very hardly depends on several factors:
- Quality of the image. The probability of faultless work of system if the person whom we try to identify, looks not directly in the chamber considerably decreases or is removed at bad illumination.
- Urgency of the photo brought in a database.
- Database size.
Technologies of recognition of the person well work with standard videocameras which are transmitted by data and cope the personal computer, and the permission 320×240 pikselov on inch require at speed of video of a stream, at least, 3 – 5 shots in a second. For comparison – comprehensible quality for video of conference requires speed of a video stream already of 15 shots in a second. Higher speed of a video stream at higher permission conducts to improvement of quality of identification. At recognition of persons from the big distance there is a strong dependence between quality of a videocamera and result of identification.
The volume of databases at use of standard personal computers does not exceed 10000 images.
The conclusion
Methods of recognition of persons offered today are interesting and close to wide introduction, however is not possible yet as at cinema to trust door opening only to technology of recognition on the person. It is good as the assistant for the security guard or other monitoring system of access.
This method is used in many situations when it is required to be convinced, that the presented document really belongs to the person who has presented it. It occurs, for example, at the international airport when the frontier guard verifies a photo on the passport with the person of the holder of the passport and the decision makes, its it is the passport or not. On similar algorithm the computer system of access acts also. Difference consists only that the photo is compared with already hranimyim in a database a template.
There were technologies which are based on recognition of persons in infra-red light. The new technology is based that the thermal picture created by radiation of heat by blood vessels of the person or, in another way, of thermogram of the person of the person, is unique for each and, hence, can be used as the biometric characteristic for access monitoring systems. Given thermogram the geometry of the person as does not depend almost on change of appearance of the person is stabler identifier, than.
Biometricheskaya identification in the integrated systems of safety
Traditional methods of identification of the person in which basis there are various identification cards, keys or unique data, suchas, for example, the password are not reliable in that degree which is required for today. Natural step to increase of reliability of identifiers of a steel of attempt of use of biometric technologies for safety systems.
The range of the problems, which decision can be found with use of new technologies, is extremely wide:
- To prevent penetration of malefactors on protected territories and in premises at the expense of a fake, thefts of documents, cards, passwords;
- To limit access to the information and to supply personal responsibility for its safety;
- To supply the admission to responsible objects only the certificated experts;
- To avoid an overhead charge connected with operation of monitoring systems of access (a card, keys);
- To exclude the inconveniences connected with uterey, damage or elementary zabyivaniem keys, cards, passwords;
- To organise the account of access and attendance of employees.
To working out of technologies for recognition of images under various biometric characteristics have started to attend already for a long time, the beginning was necessary in 60th years. Our compatriots have achieved significant successes in working out of theoretical bases of these technologies. However practical results are received basically in the west and only "yesterday". Capacity of modern computers and advanced algorithms have allowed to create products which under the characteristics and a parity became accessible and interesting to wide sections of users.
The idea to use individual characteristics of the person for its identification is not new. For today a number of technologies which can be involved in systems of safety for identification of the person on is known:
- To prints of fingers (both separate, and hands as a whole);
- To features (on the basis of optical and infra-red images);
- To iris of the eye of eyes;
- To voice;
- To other characteristics.
All biometric technologies have general approaches to the decision of a problem of identification though all methods differ convenience of application, accuracy of results.
Any biometric technology is applied stage by stage:
- Object scanning;
- Extraction of the individual information;
- Template formation;
- Comparison of a current template with a database.
The biometric system of recognition establishes conformity of concrete physiological or behavioural characteristics of the user to some set template. Usually biometric system consists of two modules: the module of registration and the identification module.
The registration module "trains" system to identify the concrete person. At a stage of registration the videocamera or other gauges scan the person to create digital submission of its shape. Scanning of the person lasts nearby 20 – 30 seconds therefore some images are formed. In an ideal case, these images will have slightly various foreshortenings and looks that will allow to obtain more exact data. The special program module processes this submission and defines prominent features of the person, then creates a template. There are some parts of the person which practically are not changed with the course of time, it, for example, the top outlines of eye-sockets, areas surrounding cheekbones, and mouth edges. The majority of the algorithms developed for biometric technologies, possible changes in a hairdress of the person as they do not use for the analysis of area of the person above border of growth of hair allow to consider. The template of the image of each user is stored in a database of biometric system.
The identification module receives the image of the person from a videocamera and will transform it to the same digital format in which the template is stored. The obtained data are compared with hranimyim in a database a template to define, whether there correspond these images one another. The degree of similarity required for check, represents a certain threshold which can be adjusted for various type of staff, capacities PC, time of days and of some other factors.
Identification can be executed in the form of verification, autentifikatsii or recognition. At verification identity of the received data and a template, hranimogo in a database is confirmed. Autentifikatsiya – confirms conformity of the image received from a videocamera to one of templates, stored in a database. At recognition if the received characteristics and one of hranimyih templates appear identical the system identifies the person with a corresponding template.
At use of biometric systems, especially recognition systems on the person, even at introduction of correct biometric characteristics not always the decision about autentifikatsii is true. It is connected with a number of features and, first of all, that many biometric characteristics can be changed. There is a certain degree of probability of an error of system. And at use of various technologies the error can essentially differ. For access monitoring systems at use of biometric technologies it is necessary to define, that it is more important not to pass "stranger" or to pass all “”.
The important factor for users of biometric technologies in safety systems is simplicity of use. The person which characteristics are scanned, should not test any inconveniences thus. In this plan the most interesting method is, unconditionally, technology of recognition on the person. However, in this case there are other problems connected first of all, with accuracy of work of system.
Despite obvious advantages, there is a number of negative biases against biometrii which often cause questions on, whether biometric data will be used for shadowing people and infringement of their right to a private life. Because of sensational applications and an unreasonable sensation the perception of biometric technologies sharply differs from a real state of affairs.
And still, use of biometric methods of identification has acquired a special urgency the last years. The especially sharply given problem was showed after events on September, 11th in the USA. The world community has realised degree of increase of threat of terrorism all over the world and complexity of the organisation of reliable protection by traditional methods. These tragical events have served as a starting point for attention strengthening to the modern integrated systems of safety. The opinion is well-known, that if the control at the airports was more religiously misfortunes could be avoided. And today search of other incidents guilty of a number could be essentially facilitated at use of modern systems of video observation in integration with systems of recognition of persons.
Methods of recognition of the person
Now there are four basic methods of recognition of the person:
- "eigenfaces";
- The analysis of "distinctive lines";
- The analysis on a basis of "neural networks";
- Method of "automatic processing of the image of the person".
All these methods differ with complexity of realisation and the application purpose.
"Eigenface" it is possible to transfer as "own person". This technology uses two-dimensional images in gradation grey which represent distinctive characteristics of the image of the person. A method "eigenface" are often used as a basis for other methods of recognition of the person.
Combining characteristics 100 – 120 "eigenface" it is possible to restore a considerable quantity of persons. At the moment of registration, "eigenface" each concrete person it is represented in the form of a number of factors. For a verification mode in which the image is used for check of identity, the "live" template is compared to already registered template, for the purpose of definition of factor of distinction. Distinction degree between templates also defines the identification fact. The technology "eigenface" is optimum at use in well shined premises when there is a possibility of scanning of the person en face.
Analysis technique of "distinctive lines" – most widely used technology of identification. This technology is similar to a technique "Eigenface", but in bolshey degrees is adapted for change of appearance or a mimicry of the person (smiling or frowning person). In technology of "distinctive lines” tens prominent features of various areas of the person, and with allowance for their relative site are used. The individual combination of these parametres defines features of each concrete person. The person of the person is unique, but is dynamical enough, since the person can smile, release a beard and moustaches, to put on points – all it increases complexity of procedure of identification. Thus, for example, at a smile some displacement of parts of the person possessed about a mouth is observed, that in turn will cause similar movement of adjacent parts. Considering such displacement, it is possible to identify unequivocally the person and at various mimic changes of the person. As this analysis considers local plots of the person, maximum deviations can be in limits to 25 ° in a horizontal plane, and approximately to 15 ° in a vertical plane and the powerful and expensive equipment requires enough, that accordingly reduces degree of distribution of the given method.
In a method based on a neural network, prominent features of both persons – registered and checked are compared on coincidence. "Neural networks" use the algorithm establishing conformity of unique parametres of the person of the checked person and parametres of a template, being in a database, the greatest possible number of parametres is thus applied. In process of comparison discrepancies between the person checked and a template from a database are defined, then the mechanism which by means of corresponding weight factors defines degree of conformity of the checked person to a template from a database is started. This method increases quality of identification of the person in difficult conditions.
Method of "automatic processing of the image of the person" – the most simple technology using distances and the relation of distances between easily defined points of the person, such as eyes, the nose end, mouth corners. Though the given method not so powerful as "eigenfaces" or "a neural network", it can be effectively enough used in the conditions of weak light exposure.
Recognition systems on the person, present in the market
For today a number of the commercial products intended for recognition of persons is developed. The algorithms used in these products, are various and it is for the present difficult to state an estimation what of technologies has advantages. Leaders at the moment are following systems: Visionic, Viisage and Miros.
- At the heart of appendix FaceIt of company Visionic the analysis algorithm of local signs developed at University of Rockefeller lays. One commercial company to the Great Britain integrated FaceIt into television anticriminal system under name Mandrake. This system searches for criminals on video data which act from 144 chambers united in the closed network. When identity is established, the system informs on it to the officer of safety. In Russia the representative of company Visionic is company "DanKom".
- One more leader in this area, company Viisage, uses the algorithm developed in the Massachusetsky institute of technology. The commercial companies and the state structures in many American states and in a number of other countries use company Viisage system together with identification certificates, for example, a driver’s licence.
- ZN Vision Technologies AG (Germany) offers a number of products in which the technology of recognition of persons is applied in the market. These systems are represented in the Russian market by company "Soling".
- In system of recognition of persons TrueFace of company Miros the technology of neural networks is used, and the system is applied in a complex of issue of cash of corporation Mr. Payroll also it is established in a casino and other pleasure establishments of many states of the USA.
In the USA independent experts had been held comparative testing of various technologies of recognition of persons. Results of testing are presented more low.
In practice, at use of systems of recognition of persons as a part of standard electronic security systems, the person who should be identified is supposed, that, looks directly in the chamber. Thus, the system works with rather simple two-dimensional image, that considerably simplifies algorithms and reduces intensity of calculations. But even in this case the recognition problem nevertheless is not trivial, as algorithms should consider possibility of change of level of illumination, look change, availability or absence of a make-up or points.
Reliability of work of system of recognition of persons very hardly depends on several factors:
- Quality of the image. The probability of faultless work of system if the person whom we try to identify, looks not directly in the chamber considerably decreases or is removed at bad illumination.
- Urgency of the photo brought in a database.
- Database size.
Technologies of recognition of the person well work with standard videocameras which are transmitted by data and cope the personal computer, and the permission 320×240 pikselov on inch require at speed of video of a stream, at least, 3 – 5 shots in a second. For comparison – comprehensible quality for video of conference requires speed of a video stream already of 15 shots in a second. Higher speed of a video stream at higher permission conducts to improvement of quality of identification. At recognition of persons from the big distance there is a strong dependence between quality of a videocamera and result of identification.
The volume of databases at use of standard personal computers does not exceed 10000 images.
The conclusion
Methods of recognition of persons offered today are interesting and close to wide introduction, however is not possible yet as at cinema to trust door opening only to technology of recognition on the person. It is good as the assistant for the security guard or other monitoring system of access.
This method is used in many situations when it is required to be convinced, that the presented document really belongs to the person who has presented it. It occurs, for example, at the international airport when the frontier guard verifies a photo on the passport with the person of the holder of the passport and the decision makes, its it is the passport or not. On similar algorithm the computer system of access acts also. Difference consists only that the photo is compared with already hranimyim in a database a template.
There were technologies which are based on recognition of persons in infra-red light. The new technology is based that the thermal picture created by radiation of heat by blood vessels of the person or, in another way, of thermogram of the person of the person, is unique for each and, hence, can be used as the biometric characteristic for access monitoring systems. Given thermogram the geometry of the person as does not depend almost on change of appearance of the person is stabler identifier, than.
Tatarchenko Nikolay Valentinovich, Timoshenko Svetlana Vjacheslavovna
It is granted by the journal "Special engineering"
Date of the publication 04.07.07