A Method Combined on Multi-level Factors for Fingerprint Image Quality Estimation

It is important for improving the performance of automatic fingerprint identification system to estimate the fingerprint image quality accurately and quickly. Fingerprint image can be expressed very well in the frequency field. After selecting the gray variance, gray coherence, orientation coherence in the spatial field and the dominating spectrum energy proportion in the frequency field as the quality estimation factors, the paper acquires every estimation factor and the quality point based on it firstly obtains the fingerprint image quality value adopting the BP network. The experimental results show that the method can acquire the fingerprint image quality value correctly. We should design better fingerprint image quality estimation method and afford more correct fingerprint image quality for the subsequence fingerprint image process algorithm.