Application Research On Several Algorithms For Fingerprint Image Preprocessing

Fingerprint identification technology is applied widely in various fields now, also is well known and accepted by people. Human have been studying on fingerprint identification science for 60 years, and have achieved fruitful results constantly. Many scientific research institution study on the key technology of fingerprint identification deeply, improve the traditional method, put forward a new theory,in order to obtain a more accurate recognition rate, faster recognition speed and higher security. But with the fast development of society, human’s demand for fingerprint identification is increasing all the time constantly, fingerprint identification still have many problems and defects.Therefore, to study on the key technology of fingerprint identification is still an important topic.To aim at the problem that low quality fingerprint images are easy to produce when grabbing image, this paper is mainly studying on the fingerprint image preprocessing technique to improve the quality of fingerprint image, which is including the compute of fingerprint ridge orientation and frequency, filtering and enhancement technique,binarization, thinning and extraction technology of feature.Reasearch has been conduted on the following aspects:(1)Introduces the research background, the significance of research and the present situation and achievement of research by the domestic and foreign researchers.(2)Introduce the process of fingerprint image preprocessing algorithm, analyze the calculation method of block orientation and frequency image in detail. The block orientation image is calculated based on the gradient vector, and the direction data is regulated in a certain range of 0 to 180°. Then apply a low pass filter to filter the orientation image smoothly to ensure the accuracy. Gray scale information is counted to calculate the block frequency image.(3)Make indepth study on enhancement technique of low quality fingerprint image, introduce the traditional Gabor filtering. Aiming at the shortcomings of the traditional Gabor filter, this paper propose a method about decomposition and reconstruction of fingerprint image, use a Log-Gabor filter to decompose the fingerprint image into several direction, and reconstruct into a new image, then apply a Gabor filter to enhance the image. By realizing the new method with Matlab, we can summarize that it can improve the clarity of ridge structure and restore the details in the corrosive area. By contrast, the new method is bertter than the traditonal Gabor filtering method.(4)Introduce the classical binarization and thinning algorithm-OTSU and OPTA. Make indepth study on the feature extraction technology. Due to the exist of many false feature, this paper filter them in two steps:first, propose a method to remove the false feature on the edge of fingerprint. That is calculating the sum of gray level in the eight neighborhood about the target, to determine which is a false feature point. Then remove the false feature in the center of fingerprint based on the spatial characteristics of feature. In this paper, the basic theory of matching is expounded simply. A fingerprint identification demo system is also designed with Matlab.