Moving Target Detection And Its Position Information Display In Complex Scenes

As a research hotspot in the field of Computer Vision(CV)and Digital Image Processing,the moving object detection technology has become more and more mature with the continuous efforts and innovations of researchers,and has been widely used in many fields such as video surveillance,intelligent transportation and military defense,and so on.However,in practical applications,changes of the environment and the presence of disturbances will cause that background is not absolutely static,while most of the current detection techniques can not meet these challenges well;at the same time,researchers often want to know other motion information when observing targets.These are important directions of future research on moving object detection technology.Firstly,the background,significance,research status of moving object detection technology in complex scenes and the main work are elaborated in this paper.Then,the infrared/visible image acquisition and processing system based on FPGA is introduced in detail,including its design principle and hardware components.Next,the basic principles of several common moving object detection algorithms and Local Binary Pattern operator are summarized.To solve the problems that the original SACON algorithm can not cope with sudden change of illumination,a large number of holes and discrete noise in detection results,a moving object detection algorithm combining Sample Consensus model and Local Binary Pattern operator was proposed.And after being compared with other common algorithms by simulation,quantitative analysis is given based on the evaluation index.Finally,the design of moving object detection and its azimuth information display system in complex scenes based on FPGA is explained,including hardware experimental environment,system construction,the principle of azimuth information acquisition and the design of each module,and several groups of experiments are carried out to verify the detection effect.The results show that the moving object detection system based on Sample Consensus model and Local Binary Pattern operator can adapt to the sudden change of illumination and avoid the interference of background interferers,which is adaptive to complex scenes;and the holes in the target are filled up so that the integrity is higher;at the same time,the azimuth data of the target is superimposed,which can help researchers understand the motion information of the target more comprehensively.