Research On Liveness Detection Technique In Fingerprint Identification System

With the wide application of the fingerprint identification system, people increasingly pay close attention to the fingerprint recognition system’s security and privacy protection. Although the fingerprint detection technology can be used to identify fake fingerprint, but the high reliability of the fingerprint sensor and the algorithm are very high, and the module size is large, so that the application field is limited. At present, the fingerprint sensor on the market mainstream cannot really have the fingerprint liveness detection capability. At present, the fingerprint sensor on the market mainstream cannot really have the fingerprint detection capability.We examine the lack of the fingerprint identification system, then attack experiment by the false fingerprint, and then put forward some improved methods, finally show that the proposed method is effective by results of the experiment. The main contributions as follows.1.The analysis and research of fingerprint identification system. We analyze the deficiency of the fingerprint identification system. The paper puts forward some improved method based on system level, such as adding the image quality evaluation techniques and the liveness detection technology.2 Study on false fingerprint attack. We analyze the work principle of the common fingerprint scanner, collect fingerprint film materials and fingerprint mold materials. In order to make the fingerprint film closer to the true fingerprint, the water and the collagen were used as the main liquid to dilute the fingerprint film. According to the different proportion to produce fingerprint film. The test is divided into three categories. In the test category one, we use the true fingerprint registration, and then simulate the difficult fingerprint for identification fingerprint; In the test category two, we try to fake fingerprint registration, and statistics success rate, and then the false fingerprint authentication; In the test categories three, we use ture fingerprint registration and fingerprint film attack.3. Research on image segmentation and quality evaluation of fingerprint identification system. According to the characteristics of fingerprint image, an adaptive threshold segmentation method combined with genetic algorithm and local threshold method is proposed based on the study of some segmentation algorithms. In the stage of fingerprint image quality evaluation, we have evaluated image quality by a single evaluation factor respectively then according to the level of misclassification rate to find out the best combination, The quality evaluation of the images collected by the combination of this kind of false fingerprint attack test finally.Low quality fingerprint image is excluded, and the high quality fingerprint images get into the next step in liveness detection to judge true or false fingerprints further.4.The research of liveness fingerprint detection. In view of the difference of noise and texture elements between true and false fingerprint, this chapter puts forward the combination of the wavelet analysis and multi texture in liveness detection technology. In this paper, we first extract the characteristics of wavelet decomposition coefficients of the fingerprint image., then extract GLCM、LBP texture feature from the reconstructed image. Wavelet coefficient features and texture features are combined to form feature vector, and features are selected by SFFS. An integration of SVM and KNN distinguishes the true or false fingerprint again.Without adding any hardware, this paper proposes a hierarchical method for the liveness detection of a fingerprint image quality evaluation and detection methods of false fingerprint, which can improve the success rate of the liveness detection, but also can ensure the good performance of fingerprint recognition system.