Abstract: For the feature extraction and matching algorithm to recognize the vein inside the back of human hand, a matching method combining generality with particularity is proposed. In the generality phase the characteristics of image feature descriptors are analyzed, involving the global information on vein images to determine the shape descriptors for boundary/region of which the translation, rotation, scaling and affinity are all invariable. And in the particularity phase based on wavelet moment the local features are considered to make better the shape descriptor for its higher resolution capability, with the advantages of wavelet moment taken because of its multi-scale analysis and invariant moment of wavelet. Testing results showed the highly improved recognition rate.