Research On Fingerprint Image Segmentation Based On Mutation Signal Analysis

Biological recognition technology is the technology which identity authenticationthrough the biological characteristics. It is mainly through the use of humanphysiological or behavioral characteristics to check or confirm the identity of oneperson. With the development of information technology, the application field ofbiological recognition technology become more and more widely. Among manybiometric technologies, fingerprint recognition technology became the most popularmethod of identification these technologies, which due to its uniqueness. Beforeidentification, we need to extract the entire fingerprint from the picture, and perform thesubsequent processing. However, the operation of fingerprint recognition is the verycareful procedure, which should reach the accuracy of pixel level. But during theprocess of collecting fingerprint from the fingerprint acquisition instrument, it willcause the loss of image quality seriously by the factors of the owners of the fingerprint,or the environment of the acquisition, even the noise in the picture which caused by thefingerprint catcher. For this, we should separate the prospect (the fingerprint which wewant to) from the image with noise.The purpose and significance of fingerprint segmentation is separating the effectiveinformation of fingerprint which we need from background or indistinctive fields withnoise caused by some reasons. This ensure the subsequent processing (such as featureextraction, fingerprint matched and recognized successfully) can run smoothly. Atpresent, most methods of fingerprint segmentation are sensitive to noise, and the standor fall of fingerprint segmentation algorithm affect the efficiency and accuracy of thefingerprint identification system directly. So it’s meaningful for fingerprintsegmentation field to find a method of segmentation which is not sensitive to noise. Inthis paper, the main task includes:(1)After simple overview of existing fingerprint image segmentation method, weput the mutation principle in the signal processing applied to fingerprint image. Itdetected the edge points of fingerprint-that is the mutations-by using the principle ofmutations signal analysis and wavelet transform, and marked them.(2)Using the least square method, it will curve fitting the edge of the fingerprint,extract it, and the edge will be embodied in the image to realize fingerprint imagesegmentation. The experimental results show that the new method is not sensitive tonoise. (3)By using the improved OPTA algorithm, I got the thinning one from the imagewhich has been segmented, and I refined the thinning result. Operation as follow: Itdetected the area in turn by using the remove templates. If inconformity, keep this point,or match with the reserve templates, keep it while consistent. Preliminary result showsthat there are a lot of scatter after thinned. In order to improve the thinning result, itincreased the filtration step. It filtered out the scatter with small pixel to improve thethinning result while the fingerprint effective information won’t be reduced.(4)To extract the feature points. It picked up the end points and branch points with8-neighborhood method. After further remove the false feature points, it realized theextraction of fingerprint feature points.By the final experiment, the segmentation and extraction of characteristic resultscan satisfy the follow-up identification process.