基于小波分析的指纹图像模糊边缘识别算法  

Fuzzy Edge Recognition Algorithm of Fingerprint Image Based on Wavelet Analysis

作　　者：刘博瑞 安艳 韩天红[1] LIU Bo-rui;AN Yan;HAN Tian-hong(College of Information Engineering,Tarim University,Xinjiang Alar 843300,China)

机构地区：[1]塔里木大学信息工程学院,新疆阿拉尔843300

出　　处：《计算机仿真》2021年第11期168-172,共5页Computer Simulation

基　　金：塔里木大学校长基金(TDSKYB1813,TDZKSS201913)。

摘　　要：为了平衡指纹图像识别方法的识别效果与抗噪性能,研究基于小波分析的指纹图像模糊边缘识别算法,提升指纹图像模糊边缘识别的正确率和抗噪性能。采用小波变换增强指纹图像,提高图像质量。并将图像方向场矩阵中被标记次数最多处视为图像中心点位置,以此为中心切割指纹图像,提取指纹有效区域。采用3阶Haar小波分解,提取有效区域内指纹图像边缘特征向量,将其作为概率神经网络(PNN)的输入。通过PNN的分类器完成指纹图像模糊边缘的某一特征向量和其它特征向量的分类和识别。实验结果表明:上述算法的指纹图像模糊边缘识别正确率为98%以上;在不同噪声环境下方法的抗噪性能良好,品质因数均较高,且识别时间短。In order to balance the recognition effect and anti-noise performance of fingerprint image recognition methods, the fingerprint image fuzzy edge recognition algorithm based on wavelet analysis was studied to improve the accuracy and anti-noise performance of fingerprint image fuzzy edge recognition. Wavelet transform was introduced to enhance fingerprint image quality. The most marked position in the image direction field matrix was regarded as the center of the image, and used to cut the fingerprint image and extract the effective region of the fingerprint. The third-order Haar wavelet decomposition was applied to extract the feature vector of fingerprint image edge in the effective region, and used as the input of probabilistic neural network(PNN). According to the PNN classifier, the classification and recognition of one feature vector and other feature vectors of fuzzy edge of fingerprint image were completed. The experimental results show that the accuracy of fuzzy edge recognition of fingerprint image is more than 98%;In different noise environments, the method has good anti-noise performance, high quality factor and short recognition time.

关 键 词：小波分析 指纹图像 模糊边缘 有效区域 边缘特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]