Design And Realization Of Dsp-based Fingerprint Acquisition System And Fingerprint Image Preprocessing

Along with the continuous development of the global economy and information technology, security precautions have also become increasingly prominent in recent years. Operations of more and more fields rely on identification technology. Biometric technology provides significant solutions, and automated fingerprint identification (AFIS) technology is one of the best options. AFIS technology now is widely applied to various aspects, such as the immigration system, banking and other home electronic security facilities.Recently, great improvements have been achieved in AFIS techniques including fingerprint acquisition, preprocessing and matching. However, there are still some challenging tasks further lucubrated in fingerprint recognition field,e.g. image acquisition, preprocessing of low quality images, and so on. These issues seriously affect the AFIS reliability. This paper profoundly discussed these important and challenging issues.The target of this paper is to present the designing of a fingerprint acquisition system based on the DSP (TMS320VC5416) and AES2510. The experimental results showed that the image textures by the system are clearer, the background noise is smaller and effective area of a image is larger. So the system can satisfy the requirements of the fingerprint identification systems.Owing to the conditions of finger skin, the way of pressing and system itself, the captured images are sometimes degraded. Therefore, enhancing the adaptability of the fingerprint acquisition system through effective preprocessing algorithm is often crucial.The main work of this dissertation is as following:①Introduced details of the characteristics of a new swipe fingerprint sensor AES2510, as well as its operating principles, structure features. The design and implementation of a fingerprint capture system, which is based on the new swipe sensor AES2510 and a high performance DSP, are also introduced. The paper presented the hardware connection of sensor AES2510 and main processor C5416, and the peripheral circuit designs in detail.②The paper proposed a compound method combined the methods of directional graph, gray variance and long-running code. The experiments proved that it can effectively extract effective regions of different types of fingerprint images. ③Based on the investigation on Gabor filter which is a typical fingerprint image enhancement algorithm based on direction field, the paper proposed an improved Gabor filter based on the region mask. The experimental results showed that the enhancement algorithm is robust and effective, and implemtation effiiciency is impoved as well.④This paper brings forward a kind of dynamic adaptive local threshold binary algorithm based on the orientation information. The experimental results showed that the presented algorithm can achieve more exact binarization results, moreover the algorithm can repair the disconnected lines in the fingerprint image in a certain extent.⑤Finally, researched and improved conventional thinning algorithms.