图像子波／分类矢量量化及其在指纹识别中的应用

IMAGE WAVELET/CLASSIFIED VECTOR QUANTIZATION AND ITS APPLICATION IN FINGERPRINT RECOGNITION

提出了适用于子波变换系数的矢量量化方法——子波／分类矢量量化（Ｗ／ＣＶＱ）．以子波变换系数中对应于图像边缘信息的零交叉片段的具体情况为依据，将子波变换图像矢量分为不同的类别，然后为各类矢量建立码本．初步实验结果表明，该方法能克服传统矢量量化编码子波变换系数不能准确恢复有关边缘信息的缺点，此结果在光学实时指纹识别系统中的应用表明，它能够充分保留原始图像的信息。

A novel Vector Quantizer (VQ),which is suited for Wavelet Transform (WT) coefficients,named Wavelet/Classified Vector Quantizer (W/CVQ) is proposed.It is proved that the information related to image edges is the zero crossing segment in image WT result.According to zero crossing segments in vectors,vectors can be classified into several classes,and a subcode book corresponding to each class is generated,respectively.Primary experimental results show that W/CVQ can overcome the shortcoming of traditional VQ that information related to edges in WT coefficients can not be preserved perfectly.The applicatoin of the results to real time optical fingerprint recognition system shows that this method can preserve information enough to make sure the correctness of recognition of the system.