关键词: 指纹, 质量坪估, 算法, 非线性融合Gabor 特征, 黑像素比例

Abstract:

In order to solve the problem of quality difference in fingerprint in public security fingerprint database , a new fusion method to evaluate fingerprint quality is proposed by combining the spatial and the frequency features of a fingerprint image. In this method , the black-pixel ratio in the central area , the ring structure of Fourier spectrum and the directional Gabor features are first employed to respectively assess the fingerprint quality , and their corresponding evaluation performances are discussed. Then , according to the evaluation capacities of these features for different quality images , a nonlinear fusion approach is presented to obtain the final quality scores by weighing the nonlinear fusion of the features mentioned above. The fusion criterion is finally tested in a public security fingerprint database , the results demonstrate that the proposed nonlinear fusion algorithm is feasible in fingerprint quality evaluation.

Key words: fingerprint, quality evaluation, algorithm, nonlinear fusion, Gabor feature, black pixel ratio