Fingerprint Image Quality Assessment And Preprocessing

As an important approach to personal identification, significant progress has been made in the research and development of automatic fingerprint recognition. But the accuracy and speed of recognition is far from satisfactory to many practical circumstances with many impendent problems still remain to be solved. Upon profound studies on the preprocessing algorithm and fingerprint image quality assessment (FIQA) of fingerprint image, a promising resolution is proposed in this essay.First of all, upon the analysis of the commonly used FIQA method, a new evaluation algorithm are proposed in the essay. The new method, which integrates frequency domain and space domain parameter, can effectively fix the defect of each singly used parameter which results in incomplete image quality assessment. The integrated algorithm classifies the regional image evaluation into 3 grades, which solves the problem that the same images have different qualities reflected from different evaluation index. The image quality is eventually expressed by the proportion of the different-quality image blocks to the total image blocks. The experimental results shows that this algorithm can rationally distinguish fingerprint images of different qualities.Secondly, each step of the image pre processing is studies in this essay, a 2-graded algorithm is adopted in image segmentation, it can fix up the shortcomings of single segmentation algorithm which is incomplete. A new direction field acquisition algorithm is proposed concerning the relevance between direction and frequency filed, as a result, the range of the direction field acquisition is defined in a very small circle region. This can not only decreases significantly the complexity of the algorithm but also improves the accuracy of the direction acquisition. In this thesis, regional texture curvature is reckoned through calculating the distribution range of texture spectrum, and a relatively accurate calculating result is achieved, meanwhile, it provides a gist for the adjustment of the filter's angle and width. In addition, a brand-new filter is devised in this essay. This new filter can effectively prevent the valid information from being undermined. It is indicated in the experiment that the filter can evidently improve the quality of original fingerprint image and that the effect of filtering is obvious. At last, the author made some amelioration on the direction-based image binarization approach, the ameliorated method has a good effect on resisting the noise, and it can acquire a more precise threshold for binarization.