Because the orientation of each seal is not exactly the same, different focus will bring different degrees of deformation, and there are a lot of fuzzy fingerprints. How to extract features correctly and realize correct matching is the key of fingerprint identification technology. Fingerprint identification technology involves image processing, pattern recognition, machine learning, computer vision, mathematical morphology, wavelet analysis and many other disciplines.
Fingerprint identification system is a typical pattern recognition system, including fingerprint image acquisition, processing, feature extraction and comparison modules.
Fingerprint image collection: Living fingerprint images can be collected by a special fingerprint collector. At present, there are mainly living optical, capacitive and pressure-sensitive fingerprint collectors. For technical indicators such as resolution and collection area, the public security industry has formed international and domestic standards, but others still lack unified standards. According to the collected fingerprint area, it can be roughly divided into rolling fingerprint and plane fingerprint, and rolling fingerprint is widely used in public security industry. In addition, fingerprint images can also be obtained by scanners, digital cameras and so on.
Fingerprint image compression: A large-capacity fingerprint database must be compressed and stored to reduce storage space. The main methods include JPEG, WSQ, EZW, etc.
Fingerprint image processing: including fingerprint region detection, image quality judgment, mode and frequency estimation, image enhancement, fingerprint image binarization and thinning, etc.
Fingerprint classification: pattern is the basic classification of fingerprints, which is divided according to the central pattern and the basic pattern of triangle. Patterns are subordinate to patterns and named after the shape of the center line. China Top Ten Fingerprint Analysis Method divides fingerprints into three types and nine forms. Automatic fingerprint identification system generally divides fingerprints into arch lines (arc lines and tent lines), basket lines (left and right baskets), bucket lines and miscellaneous lines.
Fingerprint morphology and detail feature extraction: Fingerprint morphology features include center (up and down) and triangle points (left and right). The minutiae feature points of fingerprint mainly include the starting point, ending point, combining point and bifurcation point of ridge line.
Fingerprint comparison: rough matching can be carried out according to fingerprint patterns, and then accurate matching can be carried out by using fingerprint shapes and details to give similarity scores of two fingerprints. According to different applications, the similarity scores of fingerprints are sorted or the judgment results of whether they are the same fingerprint are given.
Nowadays, computer applications, including many very confidential file protection, mostly use the method of "user ID+ password" for user identity authentication and access control. However, if passwords are forgotten or stolen by others, the security of computer systems and files will be threatened.
With the progress of science and technology, fingerprint identification technology has begun to slowly enter the computer world. At present, many companies and research institutions have made great breakthroughs in the field of fingerprint identification technology and launched many application products that perfectly combine fingerprint identification with traditional IT technology. These products have been recognized by more and more users. Fingerprint identification technology is mostly used in business areas with high security requirements, while internationally renowned brands such as Fujitsu, Samsung and IBM, which have made great achievements in the field of business mobile office, have mature fingerprint identification systems in technology and application. Here is a brief introduction to the application of fingerprint identification system in notebook computers.
As we all know, two years ago, some brands of notebooks used fingerprint identification technology to identify users when logging in. However, the fingerprint system introduced at that time belongs to optical identification system, and it should belong to the first generation fingerprint identification technology according to the present statement. Optical fingerprint identification system can only scan the surface of finger skin or dead cortex, but can't go deep into the dermis.
In this case, the cleanliness of the finger surface directly affects the recognition effect. If there is a lot of dust on the user's fingers, recognition errors may occur. Moreover, if people make fingerprint hand models according to their fingers, they may also pass through the recognition system, so users are not very safe and stable to use.
Therefore, the second generation capacitive sensor appeared. Capacitive sensor technology adopts alternating instructions and parallel arrangement of sensor plates. The alternating plate is in the form of two capacitor plates, and the valleys and ridges of the fingerprint become the dielectric between the plates. A dielectric constant sensor therebetween detects the change to generate a fingerprint image. However, because the sensor surface is made of silicon material, it is easy to be damaged and its service life is shortened. The fingerprint image is formed by the concave and convex between the valleys and ridges of the fingerprint, so the recognition rate of dirty fingers, wet fingers and other difficult fingers is low.
Today, the third generation of biometric fingerprint identification technology appears. The technology of RF sensor is to control and measure the texture of the inner layer by emitting a small amount of RF signals from the sensor itself, so as to obtain the best fingerprint image. Therefore, the pass rate of dry fingers, middle fingers, dry fingers and other difficult fingers is as high as 99@%, and the anti-counterfeiting ability is strong. The recognition principle of fingerprint sensor only responds to human dermal skin, which fundamentally eliminates the problem of artificial fingerprint. Wide temperature zone: suitable for particularly cold or extremely hot areas. Because RF sensors can produce high-quality images, RF technology is the most reliable and powerful solution. In addition, high-quality images also allow the number of sensors to be reduced without sacrificing the reliability of authentication, thus reducing the cost and applying the RF sensor idea to any field with unlimited movement and size.