Fingerprint refers to the uneven lines on the front skin of the end of the finger. This section contains a lot of characteristic information, such as the overall characteristics of the pattern, pattern area, core point, triangle point and the number of lines; detailed features, endpoints, breakpoints, bifurcation points, triangle points, core points and so on are called "feature points". Two fingerprints often have the same general features, but their detailed features can not be exactly the same.
Fingerprint recognition is the only way to identify a person by relying on overall and detailed features. Fingerprint identification technology is the most mature and widely used biometric technology at present. Its main features are uniqueness, invariance and portability.
As a popular biometric technology, fingerprint identification has been widely concerned by the industry. There are various kinds of fingerprint identification devices on the market at present, but the main steps involved in the identification process are generally fingerprint image acquisition, fingerprint image preprocessing, fingerprint image feature extraction, matching and recognition. The preprocessing of fingerprint image generally includes image enhancement, binarization and thinning.
Fingerprint image acquisition: It is an important part of automatic fingerprint identification system. Image acquisition is the process of collecting fingerprint images in vivo through a special fingerprint acquisition instrument. With the development of semiconductor technology, CMOS fingerprint sensor, thermal sensor, ultrasonic sensor and other new sensors have appeared one after another. At present, the sensor technology is becoming more and more sophisticated, and the acquisition performance is constantly improving.
Preprocessing of fingerprint image: The collected fingerprint image is usually accompanied by a variety of noise, partly due to the acquisition instrument, and partly due to the state of the finger. The purpose of fingerprint image preprocessing is to remove the noise in the image, make the picture clear and the edge obvious, and turn it into a clear point-line map, so as to extract the correct fingerprint features. Fingerprint image preprocessing plays an important role in the whole fingerprint identification system, and its quality directly affects the effect of fingerprint identification.
Graphic enhancement: The most important step in fingerprint preprocessing is to enhance fingerprint image. The purpose of image enhancement is to suppress noise while enhancing the contrast of ridge-valley structure, connect broken ridges and separate sticky ridges, highlight some information in an image according to specific needs, and weaken or remove some unnecessary information. Fingerprint image enhancement algorithms are mainly based on Gabor filter and low quality fingerprint enhancement algorithm based on Fourier filter. Gabor filter enhancement is the most common fingerprint enhancement algorithm, which uses directional field image to enhance. The low quality fingerprint enhancement algorithm based on Fourier filter transforms fingerprint image enhancement from spatial domain to frequency domain through Fourier transform, and then carries out band-pass filtering and directional filtering on fingerprint image in frequency domain to enhance fingerprint image.
Binarization: After image enhancement, the ridge (ridge) part of the image is enhanced, but the intensity of the ridge is not the same, which shows the difference of gray value. The purpose of binarization is to make the gray value of ridges tend to be consistent and simplify the whole image to binary information in fingerprint recognition. On the one hand, the image information is compressed, the main information of ridges is retained, and the storage space is saved. On the other hand, a lot of adhesion can be removed to prepare for fingerprint feature extraction and matching.
Thinning: After the binarization of fingerprint image, the ridge still has a certain width, while fingerprint identification is only interested in the direction of the ridge, and does not care about its thickness. The purpose of thinning is to delete the edge pixels of fingerprint lines, so that they have only one pixel width, reduce redundant information and highlight the main features of fingerprint lines, so as to facilitate the subsequent feature extraction. When thinning, the connection, orientation and characteristic points of the lines should be kept unchanged, and the center of the lines should be kept basically unchanged.
Fingerprint image feature extraction: There are two main feature extraction methods, one is to extract features from gray image, the other is to extract features from refined binary image. The algorithm of extracting feature directly from gray image usually tracks gray fingerprint lines, and finds the position of feature and judges the type of feature according to the result of tracking. This method eliminates the complex process of fingerprint image preprocessing, but the algorithm of feature extraction is very complex, and because of the influence of noise and other factors, the feature information (location, direction, etc.) is not accurate enough. The method of extracting features from thinned binary images is relatively simple. After obtaining reliable thinned binary images, only a 3 *3 template is needed to extract the endpoints and bifurcations.
Matching and recognition: is the last step in fingerprint recognition system, and also the most important basis for evaluating the performance of the whole fingerprint identification system. Fingerprint matching is to judge whether two fingerprints come from the same finger according to the extracted fingerprint features. Feature matching is mainly the matching of minutiae. Comparing minutiae eigenvalues of new input fingerprints with minutiae eigenvalues of fingerprints stored in fingerprint database, we can find the most similar fingerprints as the output result of identification, that is, the process of fingerprint verification and identification, which is the ultimate goal of fingerprint identification system. Due to various factors, the feature templates obtained by the same fingerprint input twice are likely to be different. Therefore, as long as the details of the input fingerprint are similar to the template stored, the two fingerprints match.
Tiancheng Shengye is one of the earliest units engaged in fingerprint technology research and development in China. Its independently developed fingerprint recognition core algorithm FPI has a research and development cycle of more than ten years. After more than 12 years of stable application, the technology is very mature, and the technical indicators are in the leading position at home and abroad. Fingerprint-related products are based on FPI fingerprint identification algorithm as the core technology, including fingerprint authentication instrument, Smart FPI fingerprint authentication platform, biometric unified identity authentication platform Smart BIOS and many fingerprint management systems. Tiancheng Shengye took the lead in putting forward the concept of "fingerprint payment" in China. In 2008, Tiancheng Shengye launched the first fingerprint bank in China. It applied fingerprint technology to external customers of banks. In a sense, Tiancheng Shengye led the application process of fingerprint technology in domestic banks.
After more than ten years of accumulation and precipitation, the company has developed and designed all its products, hardware and software, and has all independent intellectual property rights. At present, Tiancheng Shengye has successfully applied for 27 patents in fingerprint. The second prize of Beijing Science and Technology Award has been awarded to solve the problem of low quality fingerprint images. At the same time, Tiancheng Shengye has been invited to join the National Information Technology Standardization Technical Committee (SAC/TC28) as the main contributor to the national standard of biometric identification technology, and participated in the formulation of the relevant standard of BioAPI national standard 1.1 fingerprint identification.
With the progress of science and technology, the continuous innovation of sensor technology and the continuous development of fingerprint identification technology, it provides a broader space for the application of fingerprint identification technology. In recent years, fingerprint identification system has been widely used to complete the identification and identification tasks, and fingerprint identification has been applied in all aspects of life.
Computer network security: fingerprint identification and traditional identity authentication are combined in computer to prevent the security threat when passwords are forgotten or stolen by others.
Financial institutions: banks and other institutions carry out employee identity authentication, ATM customer identity authentication, mobile payment security authentication, etc.
Government organs, enterprises and institutions: mainly used for authorization and identity authentication of managers, attendance, security area staff identity authentication, database access security control, etc.
Educational, medical and judicial institutions: schools control the identity of students and staff through access control; hospitals and other private information security certification, so that end-users can safely and conveniently inquire and verify relevant procedures, patients and donors identification, to prevent fraudulent donations; public security and judicial organs use fingerprint verification solutions to quickly and effectively identify dangerous crimes. Criminals, in order to ensure the safety of citizens;
Border control, identification on passports and visas of entry-exit personnel at airports and ports, and "blacklist" screening in border checks, etc.