指纹识别的小波变换算法

Fingerprint Identification Based on Wavelet Transform

指纹的唯一性和不变性使其在生物特征识别中占有重要地位,基于指纹识别系统进行身份认证的应用越来越广泛。在分析已有指纹识别算法的基础上,提出了一种利用小波变换的指纹识别算法。通过对指纹图像小波变换系数进行建模,发现其高频子带系数近似服从广义高斯分布。所用算法使用高斯分布的两个参数标识小波子带特征,并以此作为指纹标识参数进行特征匹配。算法的优点在于运行所需数据存储量小,适合实际应用。为验证算法可行性,对指纹库数据进行仿真,实验结果表明,算法计算高效,识别率高。

the uniqueness and invariability determine the importance of fingerprint in biometric identifica- tion. More and more Fingerprint identification systems combining safeguard like traditional ID cards or smart cards are applied in practical situation. In this paper, we analyze existing fingerprint identification algorithm and then propose a new fingerprint identification algorithm using wavelet transform. Extensive researches show that taking a wavelet transform of the fingerprint image, the high- frequency subband co- efficients follow the generalized Gaussian distribution （GGD）. The proposed algorithm uses two parame- ters that following GGD to model the characteristics of wavelet subband and as fingerprint signatures in feature matching. The algorithm requires less storage memory and has better practicability. To verify the feasibility, we implement simulation with fingerprint images database. The simulation results show that the algorithm is convenient, fast and accurate.