Research On Moving Target Detection And Tracking Algorithm In Intelligent Monitoring

The intelligent monitoring system is an important part of the security system.It is an efficient and highly preventive comprehensive security means,which plays an important role in many fields such as life,military and transportation.Intellectualization is the development trend of monitoring system,which is mainly reflected in the automatic analysis and processing of video without human intervention to achieve the purpose of target detection,tracking,behavior warning and prevention.In this paper,the moving target detection and tracking algorithm in the intelligent monitoring system is deeplystudied.The specific work is as follows:In terms of moving target detection,the common target detection algorithms are analyzed and compared,and the Gaussian mixture model is mainly studied.Firstly,in view of the slow convergence of the traditional Gaussian mixture model in the initialization phase,an initialization scheme based on k-means clustering algorithm is proposed,which can effectively improve the real-time performance of the Gaussian mixture model in the initialization phase.Then,the Gaussian mixture model updates the model with a fixed learning rate,which results in the phenomenon of "trailing" and"hollowing" in the process from long-term stationary to slow motion.Aiming at this situation,this paper proposes a Gaussian mixture model with adaptive learning rate based on pixel neighborhood information and inter-frame difference method.Experiments show that the improved algorithm can effectively overcome the phenomenon of "trailing" and"hollowing" caused by the use of fixed learning rate.Finally,aiming at the problem of dynamic shadow in the process of target detection,this paper combines multiple shadow features and Gaussian mixture model,and proposes a shadow suppression algorithm based on multi-feature fusion and Gaussian mixture model.In terms of moving target tracking,the principle analysis and experimental simulation of the traditional mean-shift tracking algorithm are carried out,and it is found that this algorithm will lead to tracking failure when the tracking target is in one or more situations such as occlusion,over-speed or color interference.Therefore,aiming at the phenomenon of color interference,the space feature and edge gradient direction feature are introduced on the basis of color feature,and the mean-shift tracking algorithm based on multi-feature weighted fusion is proposed.Aiming at occlusion and over-speed,mean-shift tracking algorithm based on multi-feature weighted fusion is combined with kalman filter tracking algorithm,and mean-shift tracking algorithm based on kalman filter and multi-feature weighted fusion is proposed.Experiments show that the improved algorithm can effectively solve the tracking failure problem caused by occlusion,over-speed and color interference.For the design and implementation of the intelligent monitoring system,the improved target detection and tracking algorithm is programmed in the VS2015 development environment,and a set of intelligent monitoring system with moving target detection,tracking and abnormal behavior detection is developed by using the OpenCV visual class library and MFC interface class library.