Target Detection And Tracking Of Unmanned Surface Vehicle Based On Optical Vision

At present,the science and technology strength of our country is booming.The information construction of ocean is put in the first place in order to achieve the goal of breaking through the limitation of the first island chain and the second island chain and developing to a broader ocean.The target tracking and detection technology of the sea surface has a very important application value in the marine information construction,and it is widely applied to military intelligence,target surveillance,shipping services,and so on.Therefore,the surface target detection and tracking technology based on optical vision has important military strategic significance and engineering application value.In particular,the detection and tracking of moving targets has made unprecedented achievements.The Unmanned Surface Vehicle(USV)is a kind of small surface movement platform that can navigate autonomously in the marine environment and complete various tasks.It is often used in target detection and tracking technology.With the continuous development of science and technology,people have higher and higher requirements on the tracking ability of USV,which puts forward higher requirements on the tracking ability of algorithm.In this paper,the target detection and tracking technology of USV based on optical vision is studied in depth.The main research work is as follows:1.A sea surface visible image preprocessing method based on optical vision is proposed.Firstly,the smoothing de-noising method of visible image of water surface is introduced,including: spatial de-noising algorithm and morphological de-noising algorithm,and the de-noising effect of these methods is compared and analyzed by simulation.An improved image de-noising algorithm based on morphological weight adaptive is proposed and its validity is verified by simulation.Then,the enhancement method of visible image of water surface is introduced.Common image enhancement algorithms are introduced,including image enhancement algorithm based on gray stretch and image enhancement method based on histogram.Finally,the image stabilization method of visible image is introduced,and its validity is verified by simulation.2.The improved otsu was used for target detection.Firstly,several common target detection methods are introduced,and their detection effects are compared and analyzed by simulation.Then,the traditional genetic algorithm is improved and applied to the image segmentation method.The algorithm improves the crossover rate and mutation rate and selects otsu as the fitness function to choose the optimal segmentation threshold.Finally,the improved algorithm is applied to target detection,and the validity of the algorithm is verified by simulation.3.Target tracking is performed with an improved mean shift algorithm.Firstly,Hough transform,a common method for sea antenna detection,is introduced,and its detection effect is simulated and analyzed.Finally,the mean shift algorithm was used for target tracking,and the traditional mean shift algorithm was improved.The Kalman filter trajectory prediction was combined with the mean shift algorithm.Judge whether the optimal solution vector and the initial vector after Kalman prediction tracking meet the Bhattacharyya coefficient according to the threshold ?.The validity of the algorithm is verified by simulation.