IOMETRIC technologies occupied the role of analyzing and measuring unique biological properties of a living

Human gait is attracted as a biometric for an automated recognition system in visual surveillance and monitoring applications. This paper proposes a system of automatic gait recognition using Speeded up robust features (SURF), which achieves high recognition accuracy results. The system proposed in the paper can be mainly divided into four procedures including human motion detection and human silhouettes extraction, interested points detection, human gait features extraction and gait classification. In the experiment, four classifiers namely sparse representation based classification (SRC), local sparse representation based classification (LSRC), Meta-sample based sparse representation classification (MSRC) and Non-negative Least Squares classification (NNLS) are applied for classification and experimental results are compared with the four different classifiers. It is easy to implement and has better recognition accuracy. Experimental results conducted on own dataset with 22 subjects (11 males and 11 females) which can achieve 98.48% correct classification rate. Keywords—Biometric, LSRC; MSRC; NNLS; Silhouettes; SURF