Research On Detection And Tracking Algorithm Of Moving Object In Intelligent Monitoring

In recent years,intelligent video surveillance has gradually become a hot research direction in the field of multimedia technology.This technology combines image processing,artificial intelligence,behavioral analysis,morphological processing and other related knowledge.The purpose is to achieve the collection,extraction,understanding and analysis of effective information in monitoring video.At present,intelligent video surveillance technology has been widely used in the fields of traffic control,public security,case investigation,military and national defense and so on.It has created considerable economic value.Moving object detection and tracking is the core technology in the field of intelligent video surveillance.The accuracy,stability and timeliness of the algorithm will directly affect the performance of intelligent video surveillance system.Although the research of target detection and tracking algorithm has made great progress in recent years,there still exist some problems that affect the practicability of the algorithm.Aiming at some problems,this paper puts forward some ideas for improvement.The specific work is summarized as follows:1)Aiming at the research of moving object detection algorithm,an adaptive Vi Be algorithm based on minimum error threshold is proposed.In view of the high error rate of ViBe algorithm in complex scene,this paper dynamically adjusts the distance determination threshold in the foreground detection process according to the variance of the sample set,and effectively improves the detection rate of the foreground pixels.In view of the problem that ViBe algorithm can not quickly eliminate ghost pixels,this paper uses the minimum error method to obtain the best segmentation threshold of each frame,and uses the threshold to eliminate the ghost pixels.In order to solve the problem that the update rate of Vi Be algorithm is difficult to match the target of different motion speed,this paper uses the gradient information of the image to match the update rate of the model with the change rate of the foreground.Experiments show that the improved algorithm effectively reduces the false alarm rate in complex scenes,and the speed of ghost suppression has also been significantly improved.In addition,the detection rate of fast moving targets is also significantly improved.2)For moving object tracking algorithm,a motion estimation based CAMSHIFT algorithm is proposed.Aiming at the problem of introducing human error easily in the initialization of CAMSHIFT algorithm,this paper combines the idea of background difference in the target detection algorithm,and realizes the automatic initialization of the search frame.In view of the problem that CAMSHIFT algorithm can easily lose target when the illumination condition changes violently,this paper adds the strategy of real-time determination of the Hue component,and realizes the real-time update of the color probability model.After high speed or short occlusion,the CAMSHIFT algorithm has the problem of tracking failure.In this paper,the motion prediction mechanism is introduced with the Kalman filtering algorithm,and the possible location of the next frame of the target is predicted in the tracking process.Experimental results show that the improved CAMSHIFT algorithm in the light can still maintain a good tracking effect under the condition of sudden change.In addition,when the target moves at high speed or is temporarily sheltered,the algorithm can still maintain a stable recognition rate.