Gait Object Extraction and Recognition in Dynamic and Complex Scene Using Pulse Coupled Neural Network and Feature Fusion

This paper proposes a gait object extraction and recognition algorithm in dynamic and complex scene. The improved Pulse Coupled Neural Network (PCNN) is used to extract the gait objects. The initial gait image is employed to train the PCNN and the trained network is used to classify the images followed. After the gait object being extracted and normalized, the gait features, including Gait Energy Image (GEI), Procrustes mean shape, the Fan-Beam transform of GEI and the feature matrix are employed to recognize the gait object. The features above were fused by Euclidean Distance. The image sequences taken from public database and daily life were used in the experiment. The results showed that the method proposed in this paper is effective for dynamic and complex scene.