Markerless human gait recognition by shape and motion analysis

Gait recognition is an attractive biometric and becoming more and more important for surveillance, control area etc. This paper proposes an automatic markerless approach for human identification using static and dynamic parameters of walking gait from low-resolution video. And we also present an efficient method for roughly classify the walking direction of human. A 2D stick figure is used to represent the human body according to the body topology. First, a background subtraction is used to separate objects from background. Gait cycle is obtained by analyzing the variety of the silhouette width and height. Then, we analyze the shape characteristic and angle joint to obtain the gait feature for recognition. The multi-class SVM is used to classify the different gaits of human and the walking direction. Recognition results show this approach is simple and efficient.