Study On Pretreatment Process Of The Fingerprint Images In Police Auto Identification System

Fingerprint verification, as a personal identification method, is one of the most important techniques to distinguish the biologic characteristics of human body. The auto fingerprint identification is currently a mainstream technique of identifying the biologic characteristics, with broad applications. Currently with the fast development of information technology and computer network, the auto fingerprint identification technique has been more and more applied in daily life and work, which is drawing much attention in the academic and business circles in the world. Nowadays there is still a long way to go to mature the technology, and especially more researches about the arithmetic of some key steps are highly expected.In the field of criminal investigation, fingerprint is one of the most important witnesses, and it is the most direct and accurate way to find out and determine criminals. At present, the popularization of the auto fingerprint identification technique in the Public Security Organ has been of great benefit to our society. But in the practice of criminal investigation, the sources of fingerprints are quite complicated, and the images of the crime scene fingerprints are commonly of poor quality, which makes image pretreatment rather difficult, and performs as a bottleneck to further develop the auto fingerprint identification system. Among the current researches of the poor quality fingerprint images, few focuses on the crime scene fingerprints.Principles of the police auto identification system are briefly analyzed in the present paper, and then the pretreatment process of the fingerprint images is studied. Firstly, a multi-segmentation algorithm of a combination of the direction and variance methods has been put forward for the pretreatment of the images from different sources. Secondly, Gabor Filter is applied to enhance the quality of the fingerprint images, in order to improve the reliability of the following processes of the system. Then a new method for the quality evaluation of fingerprint images is given here based on the calculation of the image segmentation area and the direction.The experiment with Matlab has indicated that the multi-segmentation algorithm proposed in the present paper is capable of well segmenting the images of the fingerprint images from various sources, and it has successfully improved the low image contrast and reduced the severe disturbance. The image enhancement algorithm with Gabor Filte can remarkably improve the visual effect of the fingerprint images, and the quality evaluation method presented in this paper could exclude correctly the low quality images.