| Intelligent video surveillance technology is an important technology for artificial intelligent security.It can automatically analyze image sequences captured by surveillance cameras without human interference and then issue an instruction.Thus,it is widely used in military and civilian fields.Combining performance advantages of the TMS320DM6467 chip in the field of video and image processing,this paper will study video processing technology based on DM6467 video surveillance system.This paper is divided into five parts:The first part introduces the background and significance of intelligent video surveillance and then overviews the current research at home and abroad.The second part introduces some common embedded video monitoring system and the construction of system hardware platform,including the front-end detection system and the image processing system.The third part expounds the basic principles of some common moving object detection algorithms,such as:Frame difference algorithm,Gaussian mixture model algorithm,Kernel density estimation algorithm,and Fuzzy Choquet Interal algorithm.For the shortage that classical W4 can not update the background model,this paper proposed an improved algorithm based on classical W4 and frame difference.When the moving target exists in the first frame of video,the algorithm of this paper can eliminate false detection.Then the video compression and network transmission technology are introduced.In the fourth part,we utilize multiple videos to do simulation experiments with the algorithm of this paper and other commom algorithms.Qualitative analysis and quantitative analysis are carried out.Afterwards,security detection related algorithms are executed based on the results of target detection.The fifth part introduces the study and design of video processing in intelligent video surveillance,including moving object detection,security detection related detection,H.264 video compression and network transmission.We mainly introduce the algorithm transplantation on DSP side and the design of applications on ARM side.Finally we do multiple experiments on the DM6467 board.The results show that the algorithm based on classical W4 and frame difference can extract moving objects with high accuracy,strong robustness and small error and serves well for the security detection related algorithms.What’s more,the compression ratio of video encoding is good and network transmission meets the real-time. |