ABSTRACT
ABSTRACT: This paper introduces the basic concept of fingerprint matching and the development trend based on GPU computing; Discussed computation efficiency of fingerprint matching in traditional and the advantage of fingerprint matching based on GPU; The speedup ratio test between CPU and GPU fingerprint computing program, separately according to different database scale and different minutia point number. We analyzes the operation efficiency and time distribution of GPU dynamic library by using the “Profiler” software; achieving a comprehensive analysis on the bottleneck problem of GPU computing; putting forward the optimization schemes based on the existing problems.
