Research On Fuzzy Classification And Compressed Sensing Based Palmprint Recognition

With the advances in computer technology, palmprint recognition technology has become one of the research hotspots in the core areas of pattern recognition with human-computer interaction and machine learning. Palmprint identification has been widespread concerned in the industry with low aggressive, low cost, good stability and a wide range of applications. So far, although palmprint recognition technology has become more perfect, feature extraction and feature matching are still the core issues in palmprint identification process. At the stage, the less research on feature matching impacts on the effect of palmprint recognition.The main studies of this paper are that the related technologies of palmprint recognition. By reading a lot of articles about the palmprint identification, the research results of palmprint recognition algorithms in recent years are sunk in. In this paper, various types of palmprint recognition algorithms are introduced, classified and compared. For the key point of pattern matching, the palmprint recognition algorithms are proposed respectively based on fuzzy classification and compressed sensing which combine the features of (2D)2PCA. The main works of this paper are as follows:(1):A palmprint recognition algorithm based on block bi-directional two-dimensional principal component analysis (M(2D)2PCA) and fuzzy classification is proposed for palmprint identification. First, the original palmprint images are divided nonoverlapping into pieces, sub-images training set is consisted of sub-images which corresponding to the location of each image; then features are extracted from block matrix of each image with (2D)2PCA, the same method is used to obtain the feature matrix of test sample; finally, classification results with fuzzy theory are gained. Fuzzy classification is suitable for the poor distinguished classes, tolerates the ambiguity of the data. Finally, palmprint recognition experiments are operated using the Beijing Jiaotong University palmprint database. The results show that this method gives a higher recognition rate and less recognition time.(2):A palmprint recognition algorithm fused bi-directional two-dimensional principal component analysis ((2D)2PCA) and compressive sensing is proposed for palmprint identification. First, reduce the dimensionality of palmprint ranks by the bi-directional two-dimensional principal component analysis, extract the feature matrix, as overcomplete dictionary of compressive sensing algorithm. Then through the sparse representation of the overcomplete dictionary by the classification orthogonal matching pursuit algorithm (COMP) to obtain a set of optimal sparse coefficients to reconstruct each image, and finally, the classification result can be gained by comparing the test images and reconstructed images. The experimental based on Beijing Jiaotong University palmprint database, the results show that this method can get a higher palmprint recognition rate.