Implementation Of Pre-processing Algorithm On Fingerprint Identification System Of Haerbin Police Station

We usually distinguish different persons with the help of biometric identification technology according to the unduplicated features of human body. Biometric authentication determines the identity of the person mainly through examining the physiological or behavioral features of different people. Biological features used for identification must hold the following requests:(1) Uniqueness: different people have different biological features.(2) Stability: biological features could not change over time.(3) Universality: biological features must be shared by every one.(4) Collectibles: biological features could be collected.However, it is hard to find out a biological feature which holds all these characteristics at the same time in the practical application. At present, biometric authentification technologies that are used to some extent mainly include voice recognition, face recognition, iris recognition and fingerprint recognition. In these four technologies, no one is superior to another in all aspects. Different technical methods should be taken according to different occasions.In the police detection, fingerprint recognition is superior to the other techniques. Meanwhile, it is also one branch of biotechnology that is firstly adopted, as well as one of the lowest price and the most adeptest in techniques. Usually, what we can only get at the scene of the crime is criminals'fingerprint and hair.Fingerprints are the more reliable evidence left by criminals. Designing system that is suitable for the police detection can not only indentify fingerprint at the scene, but collect and store them in the database.Fingerprint recognition technology mainly involves the following functions: fingerprint image reading, feature extraction, data preservation and feature comparison. It can get fingerprint images through fingerprint recognition software. Such one-way conversion is able to covert the fingerprint into fingerprint feature data which is needed. What's more, different fingerprint can produce different feature data. Finally, we can compare these two fingerprint templates on computer to calculate the extent of their similarity, and eventually get the conclusion of their matches.Over the last several decades, researchers have invented many fingerprint identification technologies and designed a number of fingerprint identification systems. Although these systems have a quite high recognition rate, they are only tests based on their own fingerprint database rather than on a library of a uniform standard. Besides, such complicated systems as the criminal investigation system, most of the fingerprint images are poor in quality because of the internal noise and distortion. Affected by such features as reversing and stretching, sometimes, only variations of fingerprint can be obtained, which brings about lots of difficulty in the process of fingerprint recognition, even leading to false recognition. Thus, the preprocessing of fingerprint is a key part of fingerprint identification.After reading lots of relevant literatures and books as well as a great of experiments with the help of Matlab, VC++, this paper designed a framework of the entire fingerprint preprocessing and did a lot of work in the preprocessing algorithm of fingerprint image. Based on the digital image processing technology and fingerprint recognition technology, this paper designed a set of combination algorithm for the preprocessing of fingerprint recognition.This paper stated the basic procedures of the preprocessing fingerprint identification and improved the original ideal hand-pass filter, in accordance with theory of Butter Worth high-pass filter and low-pass filter. In order to obtain the best image, we improved the parameter of the algorithm based on the experimental image. In this paper, many useless background information in the image is get rid of by using the preseparation algorithms before it is dealt with.This paper achieves three methods: a mean filtering, median filtering and edge-preserving filtering and smoothed the fingerprint image respectively so as to get a smooth image. Six classical image enhancements have been realized: the histogram equalization, enhancement of automatic contrast, enhance- ment of fixed contrast, enhancement of high-pass filtering, enhance- ment of low-pass filtering and hand-pass filtering.It also realized seven classical edge detection methods: the Robet edge detector, Sobel edge detection operator, Prewitt edge detection operator, Robinson edge detection operator, Laplacian edge detection operator, Log edge detection operator and Canny edge detection operator.As the fingerprint image has direction in the local recognition, the algorithm can filter direction of the fingerprint image with the help of direction field, connecting the missing interruption and separating the adhesive ridge. Extracting the fingerprint image ridge that had been dealt with and converting the original gray-scale image into binary image; processing the fingerprint image in more detailed, serially and parallel; dividing the two-value fingerprint ridge into a single fixed width skeleton line, as well as de-nosing processing to facilitate the future extraction.It proved that the fingerprint recognition algorithm which was mentioned in this paper can get rid of the noise and enhance the fingerprint ridge, therefore, it has certain value if applied in the fingerprint identification system.The algorithm mentioned in this paper did a better job in the processing of fingerprint image treatment. Algorithm can make fractured ridge line connected , and it has a high adaptability if there is a lot of noise in the fingerprint or the image of local gray uniform is a too thick or too weak, so that fingerprint image is enhanced.