Intelligent Video Surveillance System Design Based On Motion Detection

Accurate and reliable motion detection technology plays an important role in intelligent video surveillance,although this technology has been deeply concerned and studied by more and more scholars,however,the complex background of the real environment,different light conditions,irregular movement of flying insects,angle changes and other factors have increased the difficulty of motion detection.In this paper,the hot and difficult problems in the motion detection technology are studied and analyzed,and a set of intelligent video surveillance system is designed.Related work is as follows:Firstly,this paper take the matter from the motion detection algorithm,for most of the algorithms are using the original image to carry out background model,through this method to build the background model often contains noise information,affecting the detection results.This paper introduced Gauss smoothed gradient map to construct background model,and combined Vibe algorithm with three-frame difference method to make improvements the Vibe algorithm,the improved algorithm Promotes the motion detection effect.Secondly,For moving target behavior analysis algorithm,the behavior analysis algorithm is improved by combining Kernel Slow Feature Analysis(KSFA)into 3D Convolution Neural Network(3D CNN)The method uses slow feature analysis(SFA)to extract the slowly changing features of the input image,and uses the kernel method to reduce the complexity of the algorithm.From the perspective of time and space,the motion information in the video stream is captured by 3D convolution,and convolution and down sampling operations are performed on five channels respectively.Finally,on the basis of the above algorithms,this paper designs a set of intelligent videosurveillance system for regional security problems.The system implements the following functions: streaming media services,video playback,PTZ control,video detection and alarm.