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.