Research On The Algorithm Of Detection And Recognition Of Human In Dynamic Scene

Human target detection and recognition which is the foundation and core of the intelligent video surveillance is widely applied in the home security. The human target detection and recognition in a static background has been researched by many scholars at home and abroad for years, and great achievements have been made.However, detection and recognition of the human target in the moving background caused by the camera movement is still faced with many challenges, such as incomplete extraction of the human target, low recognition rate and so on. In this paper, moving human target detection and recognition in the moving background was studied. The main research content is as follows.After studying the detection algorithms on the human motion target in the dynamic situations, an inter-frame difference method based on the motion compensation was proposed to extract the moving human target. In the algorithm,SURF was used to extract and match image feature points, and compared with Random Sample Consensus algorithm to estimate global motion parameters. After motion compensation, inter-frame difference was used to extract moving human target.Considering the shadow's influence in the process of moving human target extraction,an image shadow removal based on improved PCNN model is proposed. The median filter and morphological filtering for noise removal were applied in this paper for possible noise as well as holes in the moving human target extraction.In the paper, the characters of human target in the moving background were studied and analyzed. The algorithm of moving human recognition based on the human geometric features, hair features and skin features was proposed since moving human target cannot be effectively recognized with a single human feature.The implement of the algorithm of moving human detection and recognition in the home intelligent video surveillance system was introduced, and then the development of dynamic libraries of the moving human detection and recognition algorithm under the Linux embedded system was described in detail. Finally, this paper applied the dynamic library to the home intelligent video surveillance system.