Video Dynamic Target Processing Method Based on FPGA

As the "eyes" of the city, video surveillance technology is getting more attention. For many complex scenes in video surveillance, such as criminal detectives, illegal vehicle identification, mall anti-theft, etc., detecting moving targets has become a demand of people. In response to such a situation, this design uses an FPGA development board with ZYNQ-7000 SoC to design a detection and processing method for moving targets. The design uses the camera to capture video images and store the data in multiple VDMAs, and then sends them into the image detection module, extracts the contour of the moving target by the frame difference method, and finally stores the result for data output display and analog alarm. The IP core of the detection algorithm is realized by Vivado HLS design in computer. The two-frame difference method and the three-frame difference method are improved by adding optimization instructions and image processing methods. In the same video, the detection results of these two methods are more intuitive than the background subtraction method. At the same time, they are not affected by the surrounding environment and are not sensitive to illumination. The system finally realizes the detection speed and clear contour extraction effect in real time, and can be applied to the field of video surveillance.