面向工业人体轮廓遮挡场景的步态识别研究OA

As a dynamic biometric technology,gait recognition has unique advantages that traditional static biometric technol-ogy does not possess,and has become one of the hot topics in current academic research.In industrial scenarios,gait recognition is suitable for identifying and monitoring the identity of individuals who exhibit abnormal behavior over medium to long distances,en-suring the safety of the recognition process without physical contact.However,people in industrial environments often carry a vari-ety of complex items,which can cause obstructions to the body,especially to the upper body,changing the contour characteristics of the human body and affecting the accuracy of gait recognition.An adaptive occlusion processing gait recognition network based on dual-segment segmentation is designed to address occlusion issues in industrial environments.The network includes a primary con-tour feature extraction module,a global feature extraction unit,a multi-scale micro action extraction module,and a feature fusion module based on adaptive occlusion perception.The experimental results show that the network has good recognition accuracy for large-scale occlusion cases.