Gait Recognition Bases on the Compressed Sensing

Nowadays, there are two primary problems in the gait recognition which are the complexity of modeling and the high-dimension of feature extraction. In the light of these two problems, we propose a method that we use the CS compressed sensing (CS) Theory to extract the gait features on the basis of researching the CS theory. Based on the sparsity of the gait images, we use the projection matrix to extract the gait features to reduce the dimension of gait feature vector. Using the database provided by the Chinese Academy of Sciences Institute of Automation as testing data, we confirm the optimal dimension of the feature vectors through experiments. The performances of experiments show the effectiveness of the algorithm we proposed.