Palmprint Recognition Using Entropy Map and 2DPCA from a Single Image Per Person

The performance of two-dimension principal component analysis(2DPCA) is affected to some extent by illumination. In order to solve the problem, a new palmprint recognition algorithm based on entropy map(EM) and 2DPCA from a single image per person is developed. First, the EM of a single palmprint image is computed to eliminate the influence illumination on original grey image. Then the 2DPCA is used to reduce the dimension of the EM and extract the feature vector. Finally, the classification is implemented by the nearest neighbor classifier. The new algorithm is tested in the PolyU plmprint database. Experimental results show that compared with principal component analysis (PCA), 2DPCA, and EM+PCA, the recognition rate of the developed algorithm is the best. The highest recognition rate is 79.6%, and all the time for feature extraction and classification is 0.452s, so it has the good recognition performance.