Study On Enhancement Algorithm And Experiments Of Low Quality Fingerprint Image Based On Wavelet Transform

The fingerprint is a character of human being. There's important sense in personal identification using the fingerprint. The ridge pattern is unchanged throughout the entire life (immutability). No two fingerprints from different fingers have been found to have the same ridge pattern (uniqueness). Both the immutability and the uniqueness properties have determined the use of fingerprint matching as one of the most reliable techniques of personal identification. It is a hot field and has a very long history more than one hundred years to research the fingerprint recognition and apply it to practice since the first scientific studies of fingerprints in the mid-1800 till today. Many notable achievements have been made in this field.The process of the fingerprint recognition includes many stages such as the fingerprint acquisition, the fingerprint image enhancement, post-enhancement processing, feature extraction, minutia matching and so on. Thus, the fingerprint recognition is a hot subject combining with knowledge of the fingerprint sensor, the image processing, pattern recognition and etc. Therefore, this dissertation, from both theoretical and experimental perspectives, investigates several major problems on fingerprint acquisition, fingerprint image enhancement and post-enhancement processing. The main research achievements are as followings:A new approach of fingerprint image enhancement based on wavelet and orientation field is proposed in this paper to denoise the original image of fingerprint, at same time, segment the gray image into binarization image. This approach preserves the important features of the fingerprint ridges so that it can realize the image enhancement.After thinning following the binarization, there are always some unexpected features (the short thorn, the short ridge and the bubble) in the thinned images. After compared with the original images, the conclusion can be drawn that those are fake features that will take negative effect on the minutia matching. The fake features are always composed of the basic features such as endings and bifurcations which frequently appear in company in a certain neighborhood. Therefore, a new post-enhancement method which can effectively eliminate the fake feature point is proposed according to the character of the fake features and the statistical threshold of local pixel number. Use the new method to search the certain neighborhood of feature points for companied feature points. If the companied feature points are found in relativistic small neighborhood, they are almost fake feature points which should be eliminated. Results of plenty of experiments illustrate that the proposed method is feasible.The fingerprint image acquisition system based on FPS200 is designed and realized which includes the outer MCU circuit, series communication circuit, the computer soft of parameters setting and acquisition. The functional parameters and the outer circuit are determined by experiments in which images of different quality are captured with different parameter settings.A lot of experiments are performed based on the newly-designed fingerprint system with proposed new method. The experimental results demonstrate that the proposed wavelet-based enhancing approach can acquire clear fingerprint images, and the post processing method can make the thinned images have less false feature points. The matching ratio can overtake a higher level.