FPGA-based Moving Target Detection And Tracking System Design

The detection and tracking technology of low altitude weapons such as helicopters and unmanned aerial vehicles has become a research hotspot in the field of low altitude defense system.However,there still exist limitations in autonomous detection and real-time tracking.By introducing the moving target detection and tracking system,traditional antiaircraft weapons achieve better timeliness,intelligence and accuracy,having broad application prospects in military area and civilian area.Based on the principle of moving target detection and tracking,the key technology problems such as detection,location and matching pursuits of low altitude targets have been pointed out.Taking FPGA as core,the overall scheme,including image acquisition,storage and display functions,is proposed.In view of random noise in image acquisition,the fast median filter algorithm is introduced while the adaptive threshold based on the local pixel information is generated.In order to solve orientation limitation of traditional edge detection algorithm,the gradient calculation template has been added.Thus the image preprocessing is realized,providing guarantees for the whole system achieving stable and precise performance.The edge detection before background subtraction algorithm is designed.Image filtering and connectivity processing has conducted adopting morphological method.The moving target tracking scheme based on the improved area-based matching,which adopts centroid orientation method to extract target template,is proposed.The measured empirical value is applied to decrease the matching area of traditional matching algorithms while the SAD algorithm serves as similarity measure criterion.Outdoor and indoor Matlab simulations prove that the proposed algorithm has stronger feasibility and environmental suitability.In order to improve the probability of success matching,the camera lens zoom method is introduced to control the size of template within given range.By replacing logical operators with hierarchical adders,the window generation method during SAD match value calculation is optimized.The template updating link is designed to build a closed-loop system based on the summary of target missing factors,greatly enhanced the reliability of the system.At last,performance tests and validation experiments on the key technologies discussed in this paper are taken.In both indoor and outdoor environments,the success probability of moving target detection and tracking is better than 90%,the frame frequency is up to 25 fps,meeting the real-time and precision requirements of individual rockets.