Real-time Target Detection And Distance Display Based On Visual Background Extraction

Moving target detection has always been a hot and difficult topic in the field of computer vision and has been widely applied in military and civil fields.Nevertheless,with the increasing complexity of application environments,traditional target detection methods cannot satisfy the requirements of applications in some aspects,such as accuracy,completeness and the property of real time.Therefore,in the aspect of improving the performance of the entire computer vision system,it is crucial to research and improve the deficiencies of existing algorithms to meet the practical application requirements.The content of this paper is mainly divided into four sections: the first section introduces the research background and significance of moving target detection technology,then introduces the development and research status of target detection technology at home and abroad;The second section first introduces the basic principles of several common moving target detection algorithms and analyzes their advantages and disadvantages,then by studying the principle and deficiencies of the existing VIBE algorithm,an improved VIBE algorithm via HSV color model and edge information is proposed,which effectively solves the problems of traditional VIBE algorithm,such as ghost,shadow misjudgment,the long-term static targets gradually incorporated into background and low completeness of detection results;The third section uses MFC to design the target detection simulation interface,which completes a series of functions from video capture to detection result preservation,including video capture module,SQL Server-based video information storage module,OpenCV-based target detection module;The fourth section first introduces the hardware architecture and software development platform of the target detection system based on FPGA,then introduces the FPGA program design of the proposed detection algorithm,and completes the basic moving target detection function.In addition,based on the target detection,the target centroid is extracted and marked in real time,the distance information is superimposed and displayed so that the target information is more intuitive and richer.Finally,several experiments were carried out on the hardware platform.Results show that the proposed detection algorithm can remove ghost and shadow,effectively suppress the long-term stationary target from being incorporated into background and fill the internal holes of test results to some extent,thus improving accuracy while meeting the real-time detection requirements.