Research And Implementation Of High-definition Video Target Detection And Tracking Based On FPGA

With the rapid development of intelligence,informatization and digital application technology,people have higher and higher requirements for high-definition and real-time video image processing technology,and it is difficult to realize moving target detection and tracking algorithms purely through software.To meet actual needs,hardware implementation of video image processing technology has become a current research trend and hot spot.FPGA has the characteristics of rich logic resources,strong real-time performance,and flexible programming in the field of video image processing,so it is very suitable as a hardware platform for optimized and accelerated video image moving target detection and tracking algorithms.This paper proposes an FPGA-based real-time detection and tracking system for high-definition targets.By using the parallel mechanism and pipeline processing mechanism of FPGA,the video image processing algorithm is transplanted to the FPGA hardware platform.By making full use of the FPGA IP core and DSP computing unit,Improve the real-time performance of the high-definition video target detection and tracking system.The whole system adopts a top-down design idea,and divides the system into four modules: video image acquisition,image processing,image buffering,and image display in a modular manner.details as follows:(1)Through the research and analysis of the system's high-definition and real-time requirements,the stream processing implementation plan of the system is determined.Choose low-cost FPGA chip as the main control chip,choose the OV5640 camera to complete the collection of high-definition video with a resolution of 1080 p and a frame rate of 30 fps,choose the on-board large-capacity DDR3 to cache the video image,and use the HDMI display to detect The tracking effect is displayed,thus completing the construction of the high-definition video image processing platform.(2)Factors such as weak light and light,and object detection,and submission of one type of fusion Sobel.In the progressive grayscale calculation method,comprehensive grayscale calculation method,rapid medium-sized wave calculation method,Sobel edge detection algorithm,three-frame difference calculation method,morphological processing calculation method and the smallest rectangular frame calculation method.(3)Improving the improvement of the target,the target,the imitation of the target,the submission of the type of Camshift Kalman Arithmetic Prophecy Comprehensive quasi-conformity.(4)Utilization Verilog HDL Hard matter description language and Xilinx-like IP nucleus Xilinx efficiency approach 30 fps,actuality comparison.The last general in this system's systematic intelligence class,the actual state of the system,the systematic effectiveness,the practicality,and the practicality of the system.