Design And Implementation Of Real-time Video Surveillance System Based On Zynq

With the continuous growth of the world’s population,malignant events often occur.How to effectively carry out public security prevention has become a key concern of all countries in the world.As an important part of public security,real-time video surveillance system has been widely used in public security,transportation,military and other fields.However,in the face of the increasing amount of image information,if only through software to process video images,it will face problems such as high power consumption,high cost and slow processing speed,which is difficult to meet the needs of current video surveillance applications.To solve the above problems,this paper designs a real-time video surveillance system based on Zynq.Two traditional video surveillance moving target detection algorithms are migrated to the Zynq-7020 hardware platform according to the design idea of software and hardware cooperation,which can effectively reduce the system power consumption,hardware cost and improve the processing speed.The real-time video stream collected by the OV5640 camera can be processed.When a moving target appears in the video stream,the target can be quickly separated from the background,and the real-time detection image can be viewed through the HDMI(High Definition Multimedia Interface)display.In order to improve the picture definition and reasonably control the consumption of hardware resources,this paper improves the camera resolution to 1280×720.Considering that the real-time target detection system is difficult to pause the current image,which brings burden to the later image analysis of users,this paper designs the function of video image capture and storage.The user can capture the current video image through the key on the hardware platform,and the captured image will be directly saved to the SD(Secure Digital)memory card for later viewing.The specific design process of this paper is as follows:Firstly,the commonly used moving target detection algorithms in video surveillance are realized by MATLAB.By combining the public dataset with the test video taken by the author,the relevant algorithms are simulated and verified.After comparing the experimental results,ViBe(Visual Background Extractor)algorithm and frame difference method are selected for hardware migration.Then,in the design of the hardware part of the system,according to the characteristics of the two algorithms,this paper reasonably divides the PL(Programmable Logic)part and PS(Processing System)part of the hardware platform.By using Vivado HLS(High Level Synthesis)tool and directly using Verilog hardware description language,the core IP(Intellectual Property)cores of ViBe algorithm and frame difference method are realized respectively,and the hardware engineering of these two algorithms are built in Vivado.Finally,in Xilinx SDK(Software Development Kit),the register configuration of relevant IP cores,relevant drivers and the functional design of capturing and saving video images to SD card are realized.Through the test,it is found that the video surveillance systems of the two algorithms have achieved the expected design goals,and can effectively detect the moving targets within the range of the camera in real time.While ensuring reliability,it also has the advantages of small volume and low system power consumption,which can meet the needs of real-time video surveillance.