Scheme Realization Research On DSP Fingerprint Identification System

Fingerprint recognition plays an important role in identification of organism characters. The invariability and solitary of one fingerprint produce warranty of fingerprint as a means for identification and it is possible for it become a fact according to the development of image processing and recognition technique. Automatic fingerprint identification system (AFIS) is a technology based on computer or microprocessor with advantages of convenience and high-efficiency, with great potential.This paper is about further study and optimization of AFIS based on the research of last year. The optimization has three parts: (1) data space optimization; (2) fingerprint image pre-processing improvement; (3) matching program improvement. The RAM of DSP5402 is not large enough, so the image is divided into 4 parts, and the data space is decreased to 128kw from 256kw with the method of multiple using in different time. The pre-processing program improvement contains 3 parts, auto block size and direction smooth program, improvement of sharp filter. With these improvements, the processing effect of the fingerprint has been largely enhanced. To improve the speed of program, the optimization has three parts: (1) use mixed program of C and assemble language; (2) search the pixel with pointer; (3) use divide to replace the arc tag operation. To improve the matching program, the paper firstly gives a contrast of several usual matching methods, and then improves the feature distance match based on angle to make the matching condition automatic. The experiments show that this matching method is more precise.The whole thought of the system is depicted in this paper, and thereafter, the technique details of each parts are also described in turn. At last, the paper gives some solve methods and advice in the system debug as a reference to the AFIS based on DSP.