Research On Embedded Fingerprint Recognition System

As practical biometric recognition technologies, fingerprint recognition technology is becoming the most popular technologies of which the research institutes of home and abroad made a study. The traditional fingerprint recognition technology based on PC is restricted to home application, because of hardly mobility and high power consumption. So the embedded fingerprint recognition system with low power consumption and high performance will have bright future.In this thesis, by studying the fingerprint recognition algorithms, the algorithms are improved and innovated. An embedded fingerprint recognition system based on DSP is designed by composing the DSP technology and the improved algorithms. The system hardware platform has adopt DSP + CPLD structure, and the main hardware platform of this system consists of DSP TMS320VC5416 of TI. The systematic memory module includes procedure memory , data memory and FLASH memory. The system adopts the solid fingerprint sensor MBF200 chip of FUJITSU to acquire fingerprint image.The fingerprint recognition algorithms are designed based on this hardware platform. The fingerprint recognition algorithms includes fingerprint image preprocessing, feature extraction and feature matching. The fingerprint image preprocessing algorithm includes image segmentation, image normalizing, filtering enhancement, binarize and thinning. The image segmentation based on gray level variance separates foreground and background of the fingerprint image, and subsequently the smoothness processing accurately segments foreground and background. A 4-neighborhood smoothness filtering is added to the filtering enhancement algorithm after acquiring the direction of the image. Then the 4-neighborhood smoothing filter is done. The directional filter to enhance fingerprint image avails to processing data by the operation unit of DSP. The result of the experiment indicates that the algorithm has good effect. After studying and analyzing the relative merits of the fast thinning algorithm and OPTA thinning algorithm, the algorithm of the system adopt the composite thinning algorism which first carry out the fast thinning, and then the improved OPTA further thins the image. The result of experiment indicates that the algorithm not only has good efficient, but the skeletons of the ridges is smooth and not burr. Studying the 8-neighborhood coding ridge tracing algorithms' shortcoming and deficiency, the thesis improves and innovate the algorithm. The result of experiment indicates that improved 8-neighborhood coding ridge tracing algorithms accurately trace the ridge's tendency. The technique of bounding box is combined to match fingerprint minutia based on vector triangular.