Research Of Real-time Electronic Image Stabilization And Target Tracking Technology Based On DSP Research Of Real-time Electronic Image Stabilization And Target Tracking Technology Based On DSP

With the development of artificial intelligence,target tracking is widely used in mobile platforms,vehicles and drones,for example.It has great significance for the military and civil fields such as autonomous driving,criminal investigation tracking,and weapon guidance.In the DSP video tracking system of mobile platform,the computing and storage resources of the processor are limited,the real-time and accuracy of the algorithm are highly required.At the same time,the chattering of the platform system is serious,which has influence on the manual operation and tracking performance.Therefore,based on the research of electronic image stabilization and tracking algorithm,a real-time video image stabilization and tracking system based on DSP+ FPGA is implemented and tested.The main work of this paper includes:(1)In order to meet the need of video data acquisition and transmission in image processing system,this paper designs video decoding module in the processor to acquire Camera Link digital video on the high-speed DSP+FPGA hardware platform,utilizing the flexibility of field programming and the richness of electrical interface of FPGA.At the same time,SRIO communication module is designed by the cooperation of FPGA and DSP to realize video data transmission.After the image data is acquired by the DSP,the image algorithm is processed and transmitted to the host computer for display.The experimental results show that the video transmission speed reaches 35 frames/s.(2)In order to meet the requirement of real-time video stabilization for moving platform target tracking system,this paper chooses gray projection method which has better real-time and accuracy performance in image stabilization algorithm to stabilize image data on the basis of image acquisition.At the same time,the image stabilization algorithm is improved and optimized.The method of block matching projection is used to reduce the interference of moving objects.Mean filter is used to compensate the large translation motion of the camera and extract the camera jitter component.The experimental results on a DSP processor show that the proposed gray projection algorithm meets the real-time requirements,and the PSNR value of the optimized algorithm is 2-3 dB more than that of the traditional gray projection method.(3)In order to meet the comprehensive requirements of real-time performance and tracking performance of target tracking system,this paper uses multi-feature fusion optimization to improve particle filter algorithm,which improves the accuracy of tracking algorithm.At the same time,in order to meet the requirement of long-term tracking and adapt to the target scale change under the condition of long-term tracking,the histogram feature is used to estimate the target scale and realize the adaptive scale change.The experimental results on a DSP processor show that the algorithm has certain scale and occlusion robustness,and can adapt to long-term stable real-time tracking.In this paper,the real-time video transmission,electronic image stabilization and target tracking functions under the condition of moving platform are designed and implemented,taking full advantage of the performance advantages of the DSP processor and the FPGA processor,meeting the application requirements of video target tracking engineering.