Fingerprints are the lines formed by the concave and convex skin on the end of the finger, which (seemingly) can increase the friction force of the hand when touching objects. (In fact, fingerprints reduce the friction force and make the skin more easily stretched and deformed, which can avoid skin injury, see the section of fingerprint Uses for details). Making it easier to cling to and hold on to objects, something that evolved naturally.
There are three basic shapes of the fingerprint pattern — whorl, arch and loop. The different shapes are caused by the different directions of the pressure of the subcutaneous tissue against the top of the finger belly. Research shows that if someone points to a tall, round head, the pattern of their fingerprint will be spiral. The scientists have been able to use the model to reproduce the more common prints, as well as the less complex processes that led to the formation of rare prints.
Structure of fingerprint
Fingerprint, also known as handprint, has a broad narrow sense: narrow sense of fingerprint refers to the first section of the human finger palm skin mastoid line pattern; The broad definition of fingerprints includes lines on the fingers, knuckles and palms. There is a literal difference between fingerprint and fingerprint, that is, fingerprint refers to the mastoid line pattern on the palm skin of the first section of the finger, while fingerprint is the impression left by the mastoid line pattern. However, in judicial practice, the concept of fingerprint and fingerprint is generally universal.
Impressions are mainly caused by the existence of a large number of sweat glands and sebaceous glands on the skin of human fingers and palms (think about the sweat in your hands when you are nervous or excited). As long as life activities exist, sweat and sebum are constantly discharged, which is a bit like the ink on the surface of an atomic seal. Therefore, as long as fingers and palms touch the surface of objects, It will automatically leave an imprint like an atomic stamp. Of course, this is mainly the finger, the palm itself can leave fingerprints. If the finger or palm is stuck with other liquid substances, such as head and face grease (this is your most commonly used “mud box”), blood and ink to hold the fingerprint, the principle of leaving a fingerprint is more like a regular stamp.
Fingerprint forensics, including the search and discovery of fingerprints. Search areas for fingerprints: (1) Crime Centres; (2) the entrance and exit of the site and its surroundings; (3) articles that the criminal may have had access to; (4) Various weapons and objects left on the scene by the criminals.
And they are complex enough to provide sufficient features for identification. In addition to uniqueness, fingerprints also have heritability and invariability. It has not been found that different people have the same fingerprint, so each person’s fingerprint is different. Since fingerprints are unique to each individual, in recent centuries, the fingerprints left by criminals at the scene of crimes have become important clues for the police to hunt down suspects. Today, fingerprint identification methods have been computerized, making the identification process faster and more accurate.
Because everyone’s genes are different, so are fingerprints. However, although the formation of fingerprints is mainly influenced by genetics, there are also environmental factors (5%). Fingerprints are formed when the fetus is three to four months of development in the mother. The child’s fingerprints change slightly during development and do not become established until puberty, around the age of 14. During skin development, the epidermis, dermis, and stromal layer all grow together, but the soft subcutaneous tissue grows faster than the harder epidermis. Some people say that fingerprints can change after a bone marrow transplant, but that’s not true. Unless it’s a skin graft or an injury deep in the basal layer, the prints don’t change.
Form the basis
During the development of the skin, although the epidermis, dermis and stroma layer all grow together, the soft subcutaneous tissue grows relatively faster than the hard epidermis, so there is a constant upper pressure on the epidermis, forcing the slow-growing epidermis to shrink and collapse to the inner tissue, gradually bending and wrinkling to relieve the pressure exerted by the subcutaneous tissue. In this way, on the one hand hard to attack upward, on the other hand forced to retreat, resulting in the skin long curved, bumpy, the formation of grain. This bending and wrinkling process fluctuates as the pressure on the upper layer of the inner tissue changes, forming uneven ridges or folds until the development process stops and the final shape is a fingerprint that remains unchanged until death.
Problems of heredity
Can fingerprints be inherited?
Although no two fingerprints are identical, identical twins have similar fingerprints. Also, there are similarities in the pattern of different fingers of the same person. The study found that there were gender and ethnic differences in the types of fingerprints that appeared most frequently. For example, in the fingerprints of Chinese and Japanese, the incidence of bucket pattern and dustpan pattern is roughly the same, accounting for more than 90% of the whole. The dustpan pattern appears more frequently in European fingerprints. American fingerprints, on the other hand, have a higher incidence of arched lines.
Because the fingerprints of identical twins or people of the same race are similar, fingerprints can also be “inherited”. However, the formation of fingerprints is not only influenced by genetics, but also by environmental factors. “The details of the human body vary from person to person,” said Kaoru Inoue, a long-time anatomy professor at Hokkaido University in Japan. Fingerprints are widely used for identification because they are so easy to use.”
How are fingerprints formed?
By the time a fetus is four months old, fingerprints have formed. Before that, at about 10 weeks of development, large, bulbous bumps temporarily form in areas such as the fingertips — like the fleshy pads on a cat’s PAWS that are crucial to fingerprint formation.
When the bulge begins to shrink and collapse, the junction between the epidermis and dermis begins to fold, which is the budding “fingerprint mold”. The cells produced by the fingerprint mold are pressed hard against the surface, forming a fingerprint on the epidermis by the time the fetus is four months old. It is believed that fingerprints are developed from densely arranged “molds” on the surface of the bulge. So, the shape and size of the original bulge determines the shape of the fingerprint.
Use of fingerprint
Fingerprints are known for their “varied, lifelong” properties. Long ago, people put their fingerprints on paper or wood to identify themselves. Fingerprints have been widely used in areas such as immigration checks and the search for criminals. A fingerprint is a line of ridges and depressions on the skin, “a kind of skin stripe.” The insides of human hands and feet are covered with skin lines. Fingerprints are widely used to identify criminals, among other things. It is well known that the ancient Babylonians and the Chinese used fingerprints to verify people’s identities long ago.
Using fingerprints to identify identity, this is because fingerprints meet the following two conditions:
First of all, fingerprints are not the same. No two fingerprints are the same. Although identical twins have a high degree of similarity in pattern, their detailed features are not exactly the same, but there are some differences (see Figure 1 for characteristic points). In addition, a person’s fingerprints vary from finger to finger.
Second, fingerprints are, in principle, permanent for life. When children grow into adults, fingerprints are only enlarged and thickened, but the pattern, number and other features remain the same.
Figure 2 shows the three main types of fingerprints (dustpan pattern, bucket pattern, bow pattern) and shows a subclass of bucket pattern and bow pattern. On the right is the internal structure of the skin. Although there are other organs of touch, only those involved in the text are shown here.
Dermis is located in the lower layer of epidermis, and the junction between epidermis and dermis is uneven and complicated. These bumps are the “molds” that eventually form the fingerprint pattern. Even if the epidermis is removed, as long as the inner dermis is not damaged, the same fingerprint can still grow after healing. The sharpness of fingerprints tends to fade with age. In 1880, Folders, a British missionary and doctor living in Japan, published the first study on fingerprints in the British academic journal Nature, which scientifically explained the application of fingerprint identification in criminal investigation and other fields for the first time, and created the pioneer of modern fingerprint research, which was introduced into Asia and practiced in 1900.
In addition to helping police solve crimes, fingerprints have long been thought to enhance skin friction. But when scientists measured the effect of fingerprints on friction, they came up with a different story.
The scientists calculated the average friction caused by a fingerprint by having volunteers press their fingers against glass. The volunteers then measured the pressure on their fingers by gradually increasing it. It turns out that friction does not increase correspondingly as expected. Further microscopic examination revealed that fingerprints looked like gullies under the microscope, with Spaces between them, reducing the area of contact by about a third compared to a perfectly smooth finger surface. It’s kind of like rubber, where the friction varies with the area of contact. This leads scientists to believe that fingerprints actually reduce friction, making the skin more likely to stretch and deform, thereby preventing damage.
Method of observation
Fingerprints can be divided into three categories according to the way they are left behind:
The first category is patent print, which is visible to the eye. Such as hand paint, blood, ink and other items are printed, usually printed on the fingerprint card to become the basic information;
The second type is plastic print, which refers to fingerprints found on soft materials such as candles or clay .
The third type is latent print. This type of fingerprint is the fingerprint pattern formed by the transfer of natural secretions such as sweat, which is not visible and is the most common fingerprint in the crime scene. Latent fingerprints are often left by fingers after they come into contact with oil, sweat or dust, and then touch a clean surface. Although these fingerprints cannot be seen by the naked eye, they can be revealed by special methods and the use of some special chemical reagents.
Method of physics
If a fingerprint is left on a non-absorbent surface such as metal, plastic, glass, or tile, it is usually possible to physically make the fingerprint appear.
(1) Powder method, select the color contrast of the powder, scattered in the extraction of a complete fingerprint;
(2) Magnetic powder method, to fine iron powder particles, with a magnet as a brush, brush back and forth, show fingerprints.
(3) Laser method, with the development of laser technology, our country uses laser to show fingerprints. Argon ion laser is used as the display device. Laser can show fingerprints, that is because the human finger surface, there is always a layer of sweat and fatty acids, contact with the object will leave an inconspicuous fingerprint; A laser light, sweat, fatty acids and so on color fluorescence, fingerprints will be clear. Use a special fingerprint camera to take clear prints. After the photo is enlarged, it brings a lot of convenience to the identification work.
If fingerprints are left on absorbent surfaces such as paper, card, leather, and wood, they must be chemically treated to be visible in the laboratory.
Iodine-fumigation — that is, iodine crystals are heated to produce steam, which reacts with the grease of the fingerprint residue to produce yellow-brown fingerprints, which must be photographed immediately or fixed chemically;
Ninhydrin method — Spray the reagent on the test body and react with the amino acids of the body secretions to produce purple fingerprints;
Silver nitrate method – silver nitrate solution reacts with sodium chloride in residual sweat to produce black fingerprints in sunlight;
Fluorescent reagent method — fluorescent amines and phthalaldehyde rapidly interact with the protein or amino acid residues of fingerprints to produce a highly fluorescent fingerprint. This reagent can be used on the surface of colored objects.
Three-second gluing — cyanoacrylates are vaporized. Substances left on the fingerprint, such as amino acids and glucose, react with the vaporized gluing to make the fingerprint appear.
(1) Air observation method: Observe air on the surface of smooth objects.
(2) Naked eye observation method: with a certain light, Angle for observation. Fingerprints can often be made visible by backlighting. This is because fingerprints are often stained with dust, which absorbs light and gives them a dark color.
(3) Magnifying glass observation method: with a magnifying glass in a certain light, Angle observation.
(4) Ultraviolet observation method: with the help of ultraviolet characteristics for observation.
(5) Scientific and chemical observation method: For the potential handprints that are difficult to be observed with the naked eye, assistants should be used for chemical treatment in order to find them.
(6) Cyaniding glue method: Prepare the solution of cyaniding glue and ether first, soak the filter paper of appropriate size into the solution, take out and dry, and contact the filter paper with the fingerprint for 5~60 minutes. The part of the fingerprint near the paper is volatilized by the action of the glue, and the pattern appears on the surface of the filter paper.
A person suffering from a special disease leaves a special fingerprint at the scene, which can cause changes in the sweat of the patient. For example, if a person sweats too much and leaves fingerprints, the phenomenon of ants and bees, which is described in some novels or puzzles, is likely to occur. As mentioned on the TV some time ago, some people drink tea with inferior porcelain cups for a long time, which causes copper poisoning, resulting in the phenomenon of red sweat. A patient like this, if you leave fingerprints, they’re red.
In Argentina, fingerprint evidence was used to get a woman to confess to killing her two children, the first time modern fingerprint testing has been used in court.
With the development of science and technology, fingerprints have a new use in medicine. Some doctors have found that certain diseases can be detected by examining a person’s fingerprints or palm prints.
Fingerprints and computers have become good friends.
Many businesses are also using the characteristics of fingerprints, developed some high-tech equipment, to reflect the convenience and security of fingerprint to life, such as: fingerprint lock, fingerprint access control, fingerprint attendance machine, fingerprint acquisition instrument, fingerprint safe and network fingerprint login technology and so on. According to the survey, many high-end intelligent communities in China are equipped with fingerprint lock and fingerprint access control. The earliest device used for fingerprint is the fingerprint attendance machine. In order to put an end to punch-in, personnel managers of the company have adopted fingerprint attendance machine. At the same time, China’s first online fingerprint login technology provider has launched a test version, which is expected to solve the security problem of online accounts.
What new uses will our tiny fingerprints have in the future? A new maze lies before us, waiting to be explored, to be sought.
In addition to humans, primates such as gorillas, chimpanzees and orangutans have skin lines on their hands and feet, and even the arboreal koala and its relatives (Positridae) have skin lines, the study found. In addition, spider apes and capuchin monkeys living in South America all have a curly tail, which can be used to skillfully grasp objects, and their tails have skin stripes on the inside. It can be inferred that skin lines are always easier to form on the skin where the animal scratches the object.
Principle of recognition
Read fingerprint images, extract features, save data and compare. At the beginning, the fingerprint reading device reads the image of the human fingerprint. After the fingerprint image is taken, the original image should be preliminarily processed to make it clearer. Next, the fingerprint recognition software creates a digital representation of the fingerprint — the signature data, a single-directional conversion that can be converted from a fingerprint to a signature data but not from the signature data to a fingerprint, where two different fingerprints do not produce the same signature data.
Some algorithms combine nodes and orientation information to produce more data, which indicates the relationship between each node, while others process the entire fingerprint image. In summary, this data, often called templates, is saved as 1K sized records. No matter how they are composed, there is still no standard template, no published abstract algorithm, but individual vendors do their own thing. Finally, by the method of computer fuzzy comparison, the two fingerprint templates are compared to calculate their similarity, and finally get the matching results of the two fingerprints. Fingerprints are actually quite complicated.
Unlike manual processing, many biometrics companies do not store images of fingerprints directly. Over the years, companies and their research institutes have created a number of digital algorithms. (The relevant law in the United States considers fingerprint images to be private, so they cannot be stored directly.)
Fingerprint recognition algorithm ultimately boils down to the fingerprint image to find and compare the fingerprint features. We define two types of fingerprint features for fingerprint verification: global features and local features. General features are those that can be directly observed by the human eye, including the basic pattern loop, arch, and whorl. Other fingerprint patterns are based on these three basic patterns. It’s not enough to tell a fingerprint by pattern type alone, which is a rough classification, but it makes it easier to search for prints in large databases.
Pattern Area Pattern Area
The pattern area refers to the area on the fingerprint that includes the overall features, that is, the pattern area can distinguish the type of fingerprint. Some fingerprint recognition algorithms use only the data in the pattern area. Aetex’s fingerprint recognition algorithm uses the complete fingerprint obtained instead of just the pattern area for analysis and recognition.
The Core Point
The core point is located in the progressive center of the fingerprint grain, which is used as the reference point for reading and comparing fingerprints.
Triangular points are located at the first fork or break point from the core, or where the two lines converge, isolate, turn, or point to these singularities. The trigonometry provides the starting point for counting and tracing the fingerprint pattern.
Type Lines Type lines
A pattern line is a cross that occurs at the point where the pattern line surrounding the pattern area begins to run parallel. The pattern line usually breaks off briefly, but its outer line begins to extend continuously.
Number of fingerprint lines
(Ridge Count) Indicates the number of fingerprint ridges in the pattern area. When calculating the number of fingerprints, the core point and the triangle point are generally connected first, and the number of intersecting lines between this line and the fingerprint can be considered as the number of fingerprints. Local feature A local feature is a node on a fingerprint. Two fingerprints often share the same general features, but their local features — the Minutia Points — can never be identical. Instead of being continuous, smooth and straight, the fingerprint lines are often interrupted, split, or discounted. These breakpoints, branching points, and turning points are called “nodes.” It is these nodes that provide the fingerprint validation node feature
1. Classification – There are several types of nodes, the most typical of which are endpoints and fork points
A. Ending — where a line ends.
B. Bifurcation — where a line is bifurcated to become two or more lines.
C. Ridge Divergence — where two parallel lines draw apart.
D. DotorIsland — a line so short that it becomes a dot.
E. Enclosure — A small ring formed by dividing a grain into two and immediately merging into one is called a ring.
F. Short Ridge — a line that is short at one end but does not make a point.
2. Orientation — The nodes can move in a certain direction.
3. Curvature – Describes the speed that curvature direction changes.
4. Position — The position of a node is described by (x,y) coordinates, which can be absolute or relative to triangular points or feature points.
First, optical identification technology
Optical fingerprint collection is the oldest and most widely used technology. The finger is placed on the optical lens, the finger is illuminated by the built-in light source, the finger is projected on the charge-coupled device (CCD) with a prism, and then the ridge (the grain line with a certain width and trend in the fingerprint image) is black, the valley line (the concave part between the grain line) is white, and the digital multi-gray fingerprint image can be processed by the fingerprint device algorithm.
Two, temperature difference induction recognition technology
Its advantage is that the fingerprint image can be obtained within 0.1s, and the sensor volume and area are the smallest, that is, the commonly referred to as sliding fingerprint recognition device is the use of this technology. The disadvantage is: limited by the temperature, a long time, the finger and the chip will be in the same temperature.
Iii. Semiconductor silicon sensing technology (capacitive technology)
The semiconductor capacitance sensor can judge which position is ridge and which position is valley according to the difference between the ridge and valley of the fingerprint and the capacitance value formed by the semiconductor capacitance induction particles. It works by pre-charging capacitive sensing particles at each pixel to a reference voltage. When the finger touches the semiconductor capacitor fingerprint, because the ridge is raised and the valley is concave, different capacitance values will be formed at the ridge and valley according to the relationship between the capacitance value and the distance. The discharge current is then used to discharge. Because the corresponding capacitance value of ridge and valley is different, the discharge speed is different. The pixels under the crest (high capacitance) discharge slowly, while those under the valley (low capacitance) discharge faster. According to the different discharge rate, the position of ridge and valley can be detected, thus forming fingerprint image data.
Four, ultrasonic technology
The ultrasonic technology uses ultrasonic frequencies of 1×104Hz-1×109Hz, and the energy is controlled to the extent that it is non-destructive to the human body (the same intensity as medical diagnosis). Ultrasonic technology products can achieve the best accuracy, it has low requirements on the cleaning degree of fingers and surfaces, but its collection time will be significantly longer than the above two types of products, and the price is expensive, and can not do living fingerprint recognition, so the use of rare.
As a new IT technology field, fingerprint identification technology has many new concepts. Understanding the concept of fingerprint identification technology is helpful to accurately understand the fingerprint identification technology.
After the development of fingerprint identification system from manual identification to machine identification, it enters the automatic identification stage, which is called Automatic fingerprint identification system (AFIS). A typical automatic fingerprint identification system includes the front subsystem of interacting with people — automatic fingerprint acquisition equipment, the background subsystem of fingerprint image processing and eigenvalue extraction, and the database subsystem for fingerprint storage. When the background subsystem is used for the fingerprint registration process, it can be called the fingerprint registration subsystem. When it is used in fingerprint identification process, it is called fingerprint identification subsystem.
Fingerprint registration is also called fingerprint registration. This is the process of extracting fingerprint characteristic value from fingerprint image, forming fingerprint characteristic value template, combining with human identity information, and storing in fingerprint recognition system. It’s the equivalent of registering a fingerprint. Therefore, when registering a fingerprint, it is necessary to ensure the correct correspondence between the fingerprint and the identity information. Especially for the government, societies, companies and other units for fingerprint registration, to prevent impostor, to avoid fingerprint and identity information association error, it is very important. Therefore, in such fingerprint applications, on-site supervisors are required to participate in the fingerprint registration process. The supervisor’s fingerprint is even collected into the system as a component of the registrant’s fingerprint characteristic value template to show the importance of responsibility and provide a basis for the subsequent responsibility audit.
Verification of identification
Identification and verification is not the problem of fingerprint recognition algorithm, but the problem of fingerprint recognition system. Fingerprint recognition refers to matching fingerprint feature values in 1: N mode. It is the process of identifying a particular fingerprint from multiple fingerprint templates. The result is “yes” or “no.” Sometimes the “who” information is given.
Fingerprint verification matches fingerprint feature values in 1:1 mode. It is the process of matching the fingerprint feature template to be compared with another fingerprint feature template. The result is “no”. In a system can be used either 1:1 mode can also be used 1: N mode, which depends on the characteristics and requirements of the application system. Sometimes, the 1: N mode can be converted into 1:1 mode to improve the system security and comparison speed.
1. Fingerprints are distinct features of the human body and are complex enough to provide sufficient features for identification.
2. To increase reliability, simply register more fingerprints and identify more fingers, up to ten, each of which is different.
3. Fingerprint scanning is fast and easy to use.
4. When reading the fingerprint, the user must contact the finger with the fingerprint collecting head directly.
5. Contact is the most reliable way to read human biometrics.
6. The fingerprint collection head can be more miniaturized, and the price will be lower.
1. The fingerprint features of some people or certain groups are few and difficult to be imaged.
2. In the past, the use of fingerprints in criminal records made some people afraid to “put their fingerprints on the record.”
3. In fact, the fingerprint authentication technology can not store any data containing the fingerprint image, but only store the encrypted fingerprint feature data obtained from the fingerprint.
4. Every time a fingerprint is used, the user’s fingerprint will be left on the fingerprint collection head, and these fingerprint traces may be used to copy the fingerprint.
FRR and FAR
FRR (False Rejection Rate) and FAR (False Acceptance Rate) are the two main parameters used to evaluate the performance of fingerprint recognition algorithm. FRR and FAR are sometimes used to evaluate the performance of a fingerprint recognition system, but this is not true. In addition to the impact of fingerprint algorithm on the performance of fingerprint identification system, the impact of the performance of fingerprint acquisition equipment on the FRR and FAR should not be ignored.
FRR is commonly known as the true Rate, and the standard name is FNMR (False Non-Match rate). It can be commonly understood as the probability of “taking the fingerprint that should be matched successfully as the fingerprint that cannot be matched”. The performance measurement of fingerprint algorithm is carried out under the condition of given fingerprint database. The fingerprint database used for measurement is usually given by the organizer of the FVC (International Fingerprint Recognition Algorithm Competition). When FVC performs the performance test of fingerprint recognition algorithm, there is no external fingerprint input, so it uses the standard fingerprint image library to test. So FNMR is the test value obtained without the connection of the fingerprint collection device. The other parameters in this section are also derived from this premise.
Suppose there are 100 fingers with different ids in the fingerprint database, and each finger has three fingerprints registered, then there are 300 fingerprints in the fingerprint database. Assuming that P1 represents the ID of finger 1, its three registered fingerprints are represented by P1-F1, P1-F2, and P1-F3. FNMR “refers to the same finger 3 of the fingerprint and comparing two fingerprints, namely P1 – F1 and P1 – F2 matching, P1 – F1 and P1 – F3 matching, P1 – F2 and P1 – F3 matching, P1 – F2 and P1 – F1 match, P1 – F3 and P1 – F1 match, P1 – F3 and P1 – F2 matching, There are six matching modes. All 100 fingers were matched in 6 ways, a total of 6×100=600 matches. In theory, 600 matches can be correctly matched, and the success rate of matching is 100%. In fact, because three fingerprint images of the same finger cannot be exactly the same, there is a matching similarity problem. Suppose we set the similarity of successful matching to >90%, that is, when the similarity is greater than 90%, it means successful matching. And then we’re going to find out how many matches out of the 600 that are at least 90% similar, and that’s the number of successful matches, let’s say 570. The rest of the 600 counts represent the number of unsuccessful matches, which is 600-570=30. Then the failure rate is 30/600=5%.
For the fingerprint recognition algorithm, the matching failure rate FNMR is certain when the fingerprint database is determined. When the fingerprint library changes, its FNMR also changes. Therefore, in the world, the fingerprint database published by FVC is used as a unified test library, and the FNMR results tested in the test library are used as a standard reference to measure the performance of fingerprint algorithm.
FAR is generally known as the recognition Rate, and its standard title is FMR (False Match Rate). FMR is the most important parameter used to evaluate the performance of fingerprint recognition algorithm. It can be colloquially understood as the probability of “taking a fingerprint that should not be matched as a match”.
Take the fingerprint database in the previous segment as an example. Match each fingerprint in the library with all other fingerprints except their own, the total number of matching, that is, 300× (300-1) = 89,700 times. In theory, the number of successful matches is 6×100=600, and the number of failed matches is 89,700-600 = 89,100. It is assumed that due to the performance of fingerprint algorithm, the failed matching should be judged as successful matching, and the number of such errors is assumed to be 100 times. The error acceptance rate FAR is 100/89100=0.11%. The number of failed matches varies with the severity of the conditions that determine similarity. When the filter condition is successfully matched, that is, the threshold value increases, the FAR decreases.
FAR is also related to fingerprint databases. In the FVC competition, there are four fingerprint banks used for testing and averaging. One fingerprint database is generated manually to eliminate the influence of different fingerprint image quality caused by different acquisition equipment on the algorithm efficiency.
In the same fingerprint database, for the same algorithm, a threshold should be set as the standard to determine the similarity. If the similarity is greater than the threshold, the match is successful; otherwise, the match fails. FNMR increases with the increase of threshold value, that is, the higher the threshold value of similarity determination, the higher the probability of true fingerprint determination as false. On the contrary, FMR decreases with the increase of the threshold value, that is, with the higher the threshold value for judging the similarity, the probability of judging the false fingerprint to be true will be smaller. FAR is inversely proportional to FRR. According to the results of the 2004 FVC competition, the FNMR is generally about 5/100 when the FMR is 1/1000. In other words, if you make 1,000 matches in the fingerprint database of 100 fingers, it is possible to make one wrong match, which is a mea culpa. If 100 matches are performed, 5 matches may fail, that is, they are not recognized.
EER (Equal Error Rate) means equal error rate. This parameter is not usually used in common situations. EER is mainly used to evaluate the overall effectiveness of fingerprint algorithms. In other words, the two parameters of FAR and FRR are unified into one parameter to measure the overall performance of the fingerprint algorithm. FAR and FRR are two parameters of the same algorithm system. Put them in the same coordinate, FAR decreases with the increase of threshold value, and FRR increases with the increase of threshold value. So they must intersect. This point is the point at which FAR and FRR are equivalent under a certain threshold. It is customary to use the value of this point to measure the comprehensive performance of the algorithm. For a better fingerprint algorithm, the smaller the FAR and FRR under the same threshold, the better.
Shift both the FAR curve and the FRR curve down. And the intersection ERR is also shifted down. The smaller the EER value, the higher the overall algorithm performance.
When FRR intersects with FAR, the corresponding thresholds are very small, that is to say, the threshold of similarity at this time is not even 30%. In practice, the threshold is at least above 80%, so EER values are not commonly used to describe fingerprint algorithm performance, only in competition rankings.
FRR is actually an important indicator of the ease of use of the system. Since FRR and FAR are contradictory, there is a trade-off between ease of use and security in the design of application systems. An effective way to do this is to compare two or more fingerprints, which greatly improves system security without losing ease of use.
Boarding refusal rate
Rejection rate is rarely used and is a relatively ambiguous term in fingerprint recognition terms. In the World Fingerprint Algorithms Competition, a parameter called the rejection rate, sometimes called the denials rate, is used to measure how critical the fingerprint recognition algorithm is to the quality of the fingerprint image, using REJENROLL. That means. For a given number of fingerprints, such as 100 fingerprint images, the number of fingerprints that can be successfully registered or filed, if 99, REJENROLL=1%. For the standard fingerprint database given by FVC competition, most fingerprint algorithms can be successfully filed, that is, the REJENROLL is 0.00%.
In another context, the rejection rate is often interpreted as the probability that a fingerprint recognition system (including a fingerprint collection device) will not accept a fingerprint registration. In this case, the factors that refuse to register are more affected by the imaging capability of the fingerprint acquisition equipment besides the algorithm itself. The better the quality of the fingerprint image output by the fingerprint acquisition device, the lower the denunciation rate of the fingerprint identification system, and the lower the quality of the fingerprint image output by the fingerprint acquisition device, the higher the denunciation rate.
Registration time is another indicator used to measure the performance of fingerprint algorithms. It refers to the time from the input of fingerprint image to the success of fingerprint file (registration). According to the results of the FVC competition, the average fingerprint algorithm registration time is less than 0.5 seconds, which is one of the algorithms proposed by FVC to participate in the LIGHT group competition.
The matching time is sometimes called the matching speed, which is used to indicate the time required by the fingerprint recognition algorithm to complete a matching. It is the time calculated from the fingerprint image input to the matching result output. The matching time of most algorithms in the algorithm competition is less than 0.3 seconds. This parameter, together with the minimum registration time, constitutes the entry condition for the LIGHT group.
Because these times are affected by the quality of the fingerprint image to be tested, the average value of multiple fingerprint databases is generally taken, and the average registration time and the average matching time are generally taken as the measurement basis.
The research from “fingerprinting” to “fingerprinting” has experienced a long process. After the formation of fingerprint technology, it has undergone the development and transformation from artificial recognition technology to automatic recognition technology. With the development of computer image processing technology and information technology, fingerprint identification technology has gradually entered the field of IT technology, and many computer information systems combined together, widely used.
Japan’s National Institute of Informatics professor reminded users that when taking photos of the V-sign, it is likely to be stolen fingerprints. In addition to fingerprints, facial and iris recognition are also used in mobile phone authentication. Some administrative agencies and companies are also using the information to manage attendance. In order to obtain personal biological information, people had to get close to the person and take a photo. Recently, however, some biometric information has become available online, making the threshold for criminals much lower.
For the first time, a team from Fudan University and the Chinese Academy of Sciences has revealed that the amazing fingerprint pattern is linked to the genes of limb growth.