Research On Vision Target Tracking Algorithm Of Unmanned Surface Vehicle Based On Infrared Image

Computer vision is a science rising and developing rapidly in recent years,and target tracking technology is one of the key technologies.Target tracking has important research significance and practical application in military defense,industrial production,civil security and other aspects.The unmanned surface vehicle(USV)is a new kind of equipment applied in the marine environment,which is of great significance to the maintenance of marine sovereignty and the development of marine resources.The application of target tracking technology to surface target tracking,in cooperation with USV,will help USV to carry out many tasks,such as guard patrol,antipiracy,maritime search and rescue,obstacle avoidance navigation and hydrogeological survey.The infrared acquisition system can collect images at night and in some complex environments.The unmanned boat equipped with infrared vision platform can perform some more complex tasks in the water environment.When target tracking is applied to infrared target tracking on water surface,the actual environment of water surface should be considered.There are many problems such as target blur,jitter,occlusion and so on,which affect the target tracking effect.Based on the excellent performance of correlation filtering algorithm,this paper compares the tracking performance of different correlation filtering algorithms,including f DSST(fast Discriminative Scale Space Tracking)algorithm has good adaptability and realtime for target rotation,deformation,scale change and other problems.Combined with some problems encountered in surface unmanned boat vision infrared target tracking,the original f DSST algorithm is improved.The main research work of this paper is as follows:(1)In order to solve the problem of motion blur caused by the change of wind and wave and the motion attitude of the carrier,this paper analyzes the blur angle and blur length of the degradation function of the blurred image,and uses Wiener filter to restore the blurred image,which provides the basis of image restoration for the subsequent fuzzy image tracking.(2)In view of the problem that the tracking difficulty increases due to the uncertain offset between the frames of the vision infrared image of unmanned aerial vehicle,this paper analyzes the motion characteristics of the water antenna in the image,improves the block matching image stabilization method,and proves the effectiveness of the improvement and its shortcomings through experiments.(3)In view of the complex water environment of the unmanned boat,which is easily affected by the wind and waves and makes the position of the image taken by the vision platform suddenly change,a search scheme suitable for the vision infrared target tracking of the unmanned boat on the water is proposed.The search scheme can find the possible range of the surface infrared target in the whole image,and extract the candidate target in this range and finally pass the template match for is positioned to the target of the trace.(4)In view of the problem that the surface infrared target is occluded and lost in the field of view,the update rate and mechanism of the filter template of the f DSST algorithm are improved.The improved algorithm can better adapt to the uncertain relative motion speed between the surface unmanned boat and the target,and it will relocate the target to detect the target returning to the field of view.