Gait recognition using sparse representation

A new approach of gait recognition based on sparse representation is proposed. Static and motion information are fused using the averaged boundary which is extracted by canny operator. The training data and the testing data belong to one object when there is a sparse representation in training data for the testing data. The algorithm is implemented on USF gait database. Experimental results prove the higher performance of the method on the gait datasets which are captured on different time.