Identificazione automatizzata delle persone in base alle caratteristiche personali.
The need to identify people properly and irrevocably existed for a long time. Permission to enter a building, to open a locker, to cross a border, to withdraw money from a bank etc. is always connected to the identity of a person, so is therefore necessary to prove this identity in one way or another. We call this procedure Check. A person claiming to be entitled or have a certain identity and this must be verified. It 'a problem known to the police, for example, with people who have an identity card suspect. However, the police is often faced with another problem: who is the person who left a particular track, for example, a fingerprint, or who is this corpse. In this case, we find the identity of an unknown person, then do an identification. Specialists Biometrics use the expression one by one when it comes to verification, or one-to-many, when it comes to identification. The text that follows is essentially verifies that it is the most common case in non-related to the application of the law.
Fig. 1: Methods of verification
Probably the oldest proof of identity and authorization based on technical means, and not on personal recognition, is the mechanical key. In this case, the identity claim is based on possession. All plastic cards readable (with magnetic data storage systems, electrical or optical) are examples of the same category. Such methods of proof of authorization have reached a high technical level and some of them are very difficult to copy or forge. However, they have an inherent disadvantage: the technical system is able to verify the identity and therefore the authorization of the card or key but not the identity of the bearer. In other words, the card or key in your possession can be stolen, lost or disclosed to unauthorized persons. Knowledge-based systems rather than on possession are intended to avoid this problem. The password is the first form of this type of identification. Recently these methods have been automated through the use of passwords to computer or ID codes. It is therefore not possible improper use due to theft but it is possible that someone in some way to acquire this information and make it then misused. Despite all the precautions, a number of users noted, for example, the ID code of their credit cards, and this reduces the safety value to zero. The combination of possession and knowledge systems further reduces the possibility of misuse, but does not eliminate the main problem or that the holder is not conclusively identified. The only way to conclusively identify a person is to recognize automatically the personal characteristics. These are defined biometric features and the technology behind this identification is called Biometrics. There are many biometric characteristics that can be detected. Some of these are found in written form at any passport. However, the automated detection and comparison with automated data stored previously expected that the biometric characteristics possess the following properties:
Invariability of the properties. Must be constant for a long period of time.
Measurability. The properties must be such that it can be taken without waiting time and other complications.
Singularity. The features must be sufficiently unique properties such as to distinguish one person from any other.
Acceptability. The acquisition of these properties is possible in a manner acceptable to a large proportion of the population. Excluded are particularly invasive technologies, ie technologies involving the removal of part of the human body that would warrant an obvious disability.
Reducibility. The acquired data must be reduced to a manageable file.
Reliability. The process must ensure a high degree of reliability and reproducibility.
Privacy. The procedure does not violate the privacy of the person.
Fig. 2 Requirements for the selection of biometric properties
Given these properties, the number of biometric features used is reduced to only a few that have already been tested in the past. The table that follows provides an overview.
Features Technology Acquisition Invariance Singularity Acceptability
Geometry of the hand Optics (IR) 1:1000 Good Very good
Geometry of two fingers Optics (IR) 1:1000 Good Very good
Retina Optic Very good Not good 1:1 million (invasive)
Iris of the eye optics Very good 1:6 one million not good
Surface of the hand Optics (IR) Very good Good Not known
(pressure) Not good 1: 10000 Very good
Not good voice Electroacoustics 1: 10000 Good
Face Optical or IR Good Good Not known
Fingerprint Optical, capacitive, etc.. Very good Good 1:1 million
For a detailed discussion of the pros and cons of these technologies and state of the art, see the next chapter. Other features, such as for example the weight, the size, the color of the eyes and hair and special properties, found in passports, can not be used because it does not satisfy criteria such as the singularity, the measurability or the invariability.
Requirements for verifying the identity of a person: The person must be registered in the system as XY, and must be stored a file that contains the biometric features. Each verification begins with a recording, for example in the case of checks of fingerprints:
Fig.3 Example of the verification process
You can now take place the comparison, which shows whether the person who claims to be XY has the same biometric features. To do this you must:
Modern electronics can easily meet these requirements q. The design of these devices is based on microprocessor technology, miniaturized cameras, and other modern technology of light. The continuous decrease of the costs of the electronic components has made possible the miniaturization of devices and allowed for greater efficiency and greater economy of the unit. Some of the devices available on the market are the result of more than 15 years of development.
We can safely say that the use of biometrics technology is now mature.
Five values are important in the evaluation of biometric systems:
Although these values are considerably improved in the course of evolution of all known systems, there are still important differences between the two systems, due in part to the method of identification chosen. Up to now there is no standard for testing. Much has been done to arrive at standardization but have been encountered various difficulties due to the different nature of the systems. The value is more difficult to judge the false rejection. The false rejections depend largely on the user behavior which is why a standardization would be particularly useful.
The acquisition method most widely used today is the optical one. In most cases, are used miniature CCD cameras that detect visible light or infrared. The type of optical instrumentation depends on the detected biometric properties.
The fingerprinting involves placing the finger on a prism while the illumination is distributed in such a way as to reflect the light in its entirety on the positioning surface except for those points where the skin touches the supporting surface (reflection modified ).
The fingerprinting involves placing the finger on a prism while the illumination is distributed in such a way as to reflect the light in its entirety on the positioning surface except for those points where the skin touches the supporting surface (reflection modified ).
The acquisition of the properties of the retina uses a beam of light that scans the retina. The reflected rays produce an image in intensity of the structure of the retina detected.
The acquisition of the properties of the iris works similarly to that of the retina. The face detection captures a facial image with a camera, visible light or infrared, and processes the image to get certain criteria.
The more recent methods, in particular the acquisition of the fingerprints, they try to move away from the optical detection that requires an optical module and thus limits the miniaturization. These methods use the temperature, the pressure and / or capacity. The capacity in particular seems to be promising because it can be measured with a silicon chip miniaturized. As soon as these methods will gain precision, stability and low cost, will probably complement, if not replace, the existing optical methods. The acquisition of the signature uses a pressure-sensitive tablet or detects the position of the pen with ultrasonic or electrical systems. Speech recognition requires only a microphone of sufficient quality.
An important step in the registration process is the calculation of the template. The template, subsequently used in the process of comparison during the verification, is a data reduction of biometric features originating and should:
The more the algorithm will respond to these requirements in part contradictory, the greater will be the quality of the procedure selected. The algorithms of registration and verification are therefore to be the most important elements in biometrics. The microprocessors available 15 years ago, at the dawn of biometrics, made relatively difficult to find algorithms that were sufficiently fast and precise. Even today, many devices are connected to a high-speed PC that manages the real operation of comparison. Devices in their own right, that is independent from a PC devices such as those required by a physical access control, were, in the past, with ASICs (application-specific integrated circuits) in which the algorithm was implemented. The latest microprocessors, however, small and low power consumption, are powerful enough to handle such algorithms. Therefore, is now possible to design independent units without the cost of integration in an ASIC.
Some applications are not very demanding in terms of security of the verification (false acceptance), either because they combine various methods of verification, and because, due to their nature, do not require a high degree of safety. Other applications, particularly those intended for use in areas government, require a very high degree of safety. Test the safety of an algorithm is a difficult task. Usually a single user is not able to measure the value of False Acceptance of a given device, as it is not in possession of thousands of samples (persons) in order to obtain results of some statistical significance. The best-known suppliers of biometric products using templates database of large and sometimes publish their results. Unfortunately there are many independent institutes able to perform safety tests valid.
The geometry of the hand was one of the first methods appeared on the market. The unit called ID3I from Recognition Systems in the United States provides for the submission of the right hand and fingers are positioned as guides. The size of the hand is recorded with a camera and mirrors and, starting from such registration, is calculated a template of 9 bytes. The template is stored together with a PIN code or name of the person. The device can be used in stand-alone mode and can store up to 20,000 templates. The verification is to declare the identity of the person (eg. Entering a PIN) and the presentation of the hand, while the dimensions are compared with the template in memory. The main advantages of this device are the speed of operation, a template of small size, good acceptance by the users and the absolute non-interference with the privacy of the individual. Were, however, expressed doubts about the hygiene aspect of the method (positioning of the whole hand on the plate). Here are the characteristics of this device:
Since it was the first device with a verification time very short, was sold for various applications. However, the high value of False Acceptance (as refuted by producers) makes it not suitable for some applications. The device accepts only the right hand. The technicians who have contributed to the realization of the detector geometry of the hand, have developed a device which compares the geometry of two fingers. His name is Digi-2, and is produced in Switzerland. It 'clear that this device does not control the fingerprint but the size of the fingers. The use of this equipment is not yet widespread and features, other than those given by the manufacturer, are still unknown.
It is known that a group is working on controlling the pattern of veins on the back of the hand. The veins are recognized with an infrared camera and it is then calculated from a template. There are no known other characteristics.
A device known as “Eydentify” has been known for more than 10 years. This device performs a scan of the user's retina through a light beam and calculates a template of 256 bytes which is then used for the verification. The device has the following characteristics:
For the verification, it is necessary that between the device and the eye there is a distance of about 10 cm, which is why the positioning of the eye plays an important role. It seems that glasses and contact lenses does not affect the operation of the device. Security against counterfeiting is very high. However, this device is not very popular, because the method is not very acceptable by users.
It scans the iris of the human eye by means of a room. The iris contains a number of distinctive properties six times higher than those of the retina or fingerprint. This is therefore a method extremely reliable. The problem of the device is the positioning ie it is necessary to "scour" the user's eye. To date, only a few copies were made of such devices are not yet known and well-defined characteristics. Without doubt it is an interesting idea, but there are, as well as for the retinal scan, any doubts about the acceptability by users.
Both systems have been realized in prototypes or small series. There are no known characteristics tested. A problem is the elimination of the effects due to the change of the ambient light. Many universities are working to improve the system.
Currently, in most cases, the detection of fingerprints is performed by means of optical scanning. The finger is positioned over a prism. Where skin touches the glass, the light is scattered rather than reflected (reflection) and the resulting image is detected by a CCD camera. Have been developed other sensing devices, such as ultrasound, the thermal sensing / pressure through, for capacity. The ultrasounds have not yet been used because of the high cost and other methods are still in the test phase. The image processing and verification after the detection may be performed in two ways:
The first method is (very similar to the work of the police) to examine the so-called minutiae (terminations, grooves, bifurcations, sweat glands), which are measured and result in the template. At verification, the same procedure is used, and the result is compared with the template in memory (minutiae matching, comparison of the minutiae).
The second method stores picture elements selected as the template. At verification, these image elements are used to control the similar images presented finger and to verify that they match with the template (pattern matching, comparison of the pattern)
Both methods give similar values as a result of safety, the first method, however, requires verification times a little 'longer.
There are known several devices of this kind. The first device of this type, in chronological order, is probably the equipment Identix Inc., California, who is currently presenting the third and fourth generation of devices. The algorithm of these devices is operating in an ASIC or a microprocessor, which makes it independent of PCs connected. Other systems are Identicator (USA) (which now belongs to the group Identix), Morpho Systems (France), Startek (Taiwan), Dermoprint (Hungary), GO (Italy), Digicomp (Italy) etc.. Most of these systems has its own algorithm implemented on a PC. Not many manufacturers offer the so-called live finger detection. The purpose of this is to inhibit the verification of a copy of the finger (for example, a false silicone) or, in extreme cases, a cut finger of a registered person. Many properties can distinguish a fake from a live finger or dead, but not all are feasible because
detection is not significant enough.
He often tried to use physical properties sometimes really unusual. The following have been announced:
Shape of the outer ear: it presents difficulties when the ear is covered with hair.
Odor of the human body (!)
Structure of the palm: this property is often used by the police, but it has problems due to various reasons: the inside of the hand is curved, it is therefore difficult to detect and has a lot of information that makes it difficult selection.
These properties are not represented so far in commercially available products and we do not know the features.
The main problem in acquiring and using behavioral properties is the distinction between variables and invariable. For this reason, these properties are less exact than the physiological properties and are useful only in very special applications.
The principal attraction of such a method resides in the fact that the financial world uses the signature as the preferred method of identification. The detectors biometric signature, however, are not limited to check the image of the full signature but also the dynamics of movements during the signing. There are known various devices of this type. The value of False Acceptance is quite high (up to 10%), but it is an acceptable value for example, for applications in the banking sector, where are used in parallel with other means of identification. Many applications are not suitable because the procedure requires time and space and is useless in the case of illiterate people (developing countries).
The main advantage of speech recognition systems is that the sensor is very simple and can be placed anywhere: just up the phone. However, the values of False Acceptance and False Rejection are relatively high and this means that the method is useful only if simultaneously employ other means of identification. The devices of this type analyze the flow of energy and the spectral performance of the speech, in most cases a particular word. The devices have a high tolerance (therefore a relatively low level of security), or values of false reject high.
There have been various attempts to use the pressure exerted on the keys of keyboards of the PC as a distinguishing property. Two problems have made it difficult to approach this:
Keyboards of different brands have different characteristics
People are not accustomed to using keyboards do not normally have reproducible characteristics of its way to press.
According to our information, there are currently no products on sale with these characteristics.
Looking at the market in the last 10 years, these were found to be the most successful products: They were mainly used for detecting the hand geometry and fingerprint verification systems. It seems that the use of the hand as a means of verification is accepted by a wider audience. The verification of the retina has been used only in high-security environments and therefore has not had a major breakthrough. There have been major attempts to test iris recognition and face, still do not know the most important applications. All other methods, although interesting in special cases, have not had a significant market success.
In general, there are many possible applications for biometric systems. Their main advantage is evident in all cases in which the need is to control without any possibility of doubt the identity of a person. Because this type of identification has not yet taken steps in the market? There are several possible reasons:
The technology is relatively new. Although the first units have appeared on the market 15 years ago, these were bulky, slow and too expensive.
The evolution of the main existing systems (eg. Verification of a user of an ATM) requires considerable investment and this means that it takes time to implement the systems.
The cost of a unit biometric is still considerably higher than that of a magnetic stripe reader, although prices are decreasing rapidly.
An argument that you hear often talk about is that of non-acceptance by the users. However, experience shows that this problem is much less important than they expect the majority of potential buyers.
The lack of standardization has forced the major companies to delay the implementation of these systems because they do not want to be tied to a single vendor with a proprietary product.
However, in recent years have introduced some key applications. Here are some examples.
The earliest users of biometric systems as a means of access control to buildings and installations were various military organizations and customers with high levels of security, such as banks and nuclear power plants. More and more people realize that biometrics has advantages not only for applications with a high level of security. The simplicity of use ("the key is always with you") makes these systems very attractive for other applications. We know of various industries and service organizations that have introduced biometrics to control access not only to its employees but also to customers and visitors. We expect rapid growth in the number of applications in this field in the coming years. It will not be in any case never a very large market volume since the number of units is usually limited to the number of revenue.
A center of jewelry stores with about 5,500 employees and more than 7,000 visitors a year using fingerprint readers. More than 30 controlled access, in addition to the reception desk, have sensing devices. Several nuclear power plants integrate your system with badge access control biometric systems to protect critical areas inside. The access of trucks into a major port is protected by detectors on the geometry of the hand and, in a major airport, the truck access requires the verification of the fingerprints of the driver. The safe banks were often protected by fingerprint or face. In this way the client is able to open the safe without the intervention of the employee of the bank. In a number of banks has only just begun to control access to vault fingerprint.
Specialists believe that the fraud detection systems (stamping "contract") corresponds approximately to the loss of one hour of work per employee per week. Many employers do not accept a figure so high, anyway, fraud is still perpetuated. Particularly exposed to this type of fraud are companies with staff in frequent replacement or other temporary or seasonal. Biometrics applied to detection systems completely eliminate this kind of fraud. We calculated the examples that demonstrate that the elimination of this type of fraud has allowed us to pay off the entire system biometric 6 months. In the United States it was expected that in the near future about 10% of all detection systems will be equipped with biometric unit.
A supermarket chain with 450 stores to check out the working hours of about 7500 employees with a system of fingerprinting. This is especially recommended for a type of seasonal or replacement quickly. The units of fingerprinting must send the information via modem dial-up. The unions, who initially protested, soon convinced of the advantages presented by such a system, as the fraud perpetrated by employers (non-registration of actual working hours) is also missing. The State Policy Committee (Committee on the protection of personal data) has determined that there is no invasion of privacy.
Such applications are difficult due to the recording of a huge number of people. On the other hand, the compatibility of the various state borders is difficult to achieve with the lack of standardization of biometric systems.
Schipol airport in Amsterdam was the first airport driver to control and speed up border crossings. Frequent flyers could buy a smart card containing the owner's fingerprint template that would allow them to avoid the control of the border police arrival. The system was limited to citizens of the nation and has been deleted after a pilot phase (technically successful), as far as we can tell, for commercial reasons. Other pilot experiments were conducted at various airports in the U.S. and Canada, some using hand geometry, other fingerprints. Until now, no decisions have been made about the definitive introduction of these systems. Several countries are discussing projects to check passports and / or identity cards biometrically. So far, no country has taken final decisions are underway pilot experiments and discussion of the various projects proceeds.
The level of fraud in the payment of social security and state pension is very high in many countries. Payments are paid to deceased persons, to unauthorized persons, and there are frequent cases of double payment. In this way the state is exposed to a very substantial loss, which in some cases has reached the equivalent of the total amount of money to be paid. However it should be noted that the verification system of the aforementioned type (one to one) do not contribute to unregister repeated the same person. Therefore, the method of checking one by one should always be complemented by an accurate control (one to many) of the record in the system. Recently, there have been developed solutions much simpler and cheaper than the so-called AFIS system (Automatic Fingerprint Identification System), which are designed for police work, but at the same time less demanding because they require the same level of quality required in respect of a legal nature. The same problems have to be solved with identity cards and passports.
The first country to monitor the payments of social security with biometric systems was South Africa. In this country, even illiteracy has played a role. Other such systems are being introduced in Spain and Colombia. All three systems are based on fingerprints.
Specialists know that protect data with a password, which is the system most commonly used today, is problematic. The majority of people, unless it is unable to do so, it uses trivial passwords as their date of birth, the first name and so on. If the use of combinations trivial is excluded from the software and the system requires password changes too frequently, it develops a tendency to write somewhere in your password, which often will be on the bottom of the keyboard or in the top drawer of the desk . This reduces the security value of the password to almost zero. Be honest with yourself: you never disclosed your password over the phone to your colleague or your secretary to allow them to look at something on your PC? We know of examples in non-European countries where the password in the banking systems have resulted in massive fraud. This can potentially also in the Western world. Systems based on the card allow a greater safety and for this reason have been introduced in a number of cases also, no doubt, because of the low price of card readers. The security systems based on possession have, as already explained above, inherent disadvantages. Recently, systems based on the detection of fingerprints have gained notoriety because of problems related to the security of the data in computers. This would allow the opening of an important market for biometric systems. There are three main areas of application:
Control of access to computers. These systems protect the load on your computer, access to the operating system (such as Windows NT), or access to some directory on your hard drive. Access control for database and software on the servers they access the client. The security of these systems also requires the secure transmission between client and server.
Electronic signature. Biometric systems are particularly suitable for protecting and controlling financial transactions over computer networks. Another objective is the protection of certain transactions that take place on the internet that is extremely unsafe.
The following applications have been made known to this day. The Oracle database provider offers a system based on the detection of fingerprints for biometrically protect a database server. The biometric verification terminal is installed on the client side, the biometric information of the users, however, are stored in a special database server. At login, the user is prompted to undergo fingerprint verification. The exchange of biometric information between the client and server is protected by an encryption algorithm. A large Asian bank cashiers check transactions which must apply for approval of their supervisor whenever a transaction exceeds certain limits. This authorization is provided by a detector of fingerprints. This seems to be currently the largest biometric verification system in the world: you have installed more than 2000 units. An Egyptian bank is introducing a similar system.
Many new applications have appeared over the past two years. Biometric systems are useful in cases where it is required to obtain a reliable verification of a person.
In a country that, for political reasons, refugees in camps are not physically protected, have been equipped with smartcard containing the template of your fingerprint. They are asked to report weekly or even daily to a registration office to demonstrate its presence with smart card and fingerprint verification. Various European prisons are equipped with biometric (fingerprint and hand geometry) in order to identify visitors or check at the time of the prisoners when they leave prison for any reason, particularly in the case of persons belonging to certain groups ethnic for which the identification of the face by Europeans is difficult. A European Parliament with more than 500 delegates installed a voting system that requires fingerprint verification for each vote. This means a high level of security against fraud at the voting results.