| With the advancement and development of machine vision and UAV-related technologies,moving target detection and tracking technology has become a hot spot in current research.In UAV aerial photography,the real-time moving target detection and tracking technology of video images has been widely used,such as military reconnaissance,engineering surveying and mapping,etc.In the field of machine vision due to its large amount of information,high real-time and other advantages can provide a large amount of real-time information for the intelligent work of UAVs,the current main visual tracking system is based on GPU or DSP and other methods,not only computing overhead but also high power consumption,but also through software is gradually difficult to meet the UAV work needs in high-speed operation and real-time shooting.Based on the above background,aiming at the analysis of UAV needs for identification and real-time tracking of specific targets and attack targets,along with the application and development of embedded systems,this study adopts FPGA hardware processing to detect and track targets in aerial images,and uses bilateral filtering image preprocessing,double-frame frame difference method to detect targets and CamShift algorithm to track moving targets from the performance indicators of real-time and accuracy.The main research contents of this paper include:1.The system design adopts a low-cost FPGA chip as the core control module of the system,uses Verilog hardware description language to design in Quartus II software,uses OV5640 camera to complete video image acquisition,completes image data caching through large-capacity storage device SDRAM ping-pong operation,and finally displays video images on VGA displays to complete the construction of video image processing platform.2.In view of the random noise problem in the image acquisition process,in order to ensure the accuracy of subsequent moving target detection,the bilateral filtering algorithm is transplanted and implemented on the FPGA system development platform on the basis of mean filtering,median filtering and Gaussian filtering,and the rationality of the algorithm design is verified by joint simulation with MATLAB and evaluated by performance indicators.3.The mainstream object detection algorithm,Meanshift algorithm and CamShift algorithm are analyzed,the corresponding efficiency can be observed from the accuracy curve of each algorithm,and the inter-frame difference method is finally selected as the moving target detection algorithm of the system and the CamShift algorithm as the tracking algorithm of the system from the functional requirements of the UAV.After several tests,it is shown that the algorithm studied in this paper improves the stability,real-time and accuracy of moving target detection and tracking of the system,and finally experiments are carried out on the UAV prototype. |