Research On Target Tracking Algorithm Based On Image Restoration

Target tracking technology has always been a hotspot in the field of computer vision,and its application field is more and more extensive.As the application environment of the target tracking system is more and more complicated,the anti-jamming requirement of the target tracking technology is getting higher and higher.Under the conditions of haze weather or the presence of smoke bombs,the image obtained by image acquisition system will be serious degradation.The useful information needed by the target tracking algorithm will be reduced or even disappeared,leading to the tracking system fail to lock the target.How to effectively reduce the interference,improve the contrast of the image,recover the characteristic information of the region of interest from the disturbed image,improve the robustness of the tracking algorithm,and have certain research value.This paper first analyzes the haze image and smoke image,and then the algorithm based on polarization and local region restoration is proposed to enhance the anti-jamming of target tracking system.The algorithm makes full use of the characteristics of different interference sources.The results of this paper are as follows:Firstly,the imaging model of fog and the polarization characteristics of light are studied.The traditional polarization fog algorithm ignores the polarization characteristics of the reflected light of the object and the recovery of the mirror area is poor.In view of this shortcoming,this paper proposes a de-fog algorithm that takes into account the polarization characteristics of the reflected light of the object and the polarization degree of the atmospheric light which changes with he optical thickness,so that the information recovery in the image is more accurate.Secondly,when there is a white object in the scene,there is a large deviation in the atmospheric light value estimated by the single image de-fog algorithm based on the dark channel,which affects the image restoration effect.For this problem,we calculate the dark channel image of the best and worst polarized image,and then obtain the difference graph of two dark channel images.The atmospheric light is estimated from the difference image and the fog is removed.Thirdly,it was found that the smokes were white areas where the gray scale changed slowly by studying the images that were disturbed by smoke bombs.In view of the characteristics of smoke in the image,this paper uses the weighted fusion of the dynamic characteristics and the static features to detect the area of smoke obscure,laying the foundation for the subsequent image restoration.Fourthly,in case of the local area of the image is blocked by the smoke,the target feature in the image is masked,and the traditional image restoration algorithm based on contrast enhancement and dark channel can not recover the characteristic information from smoke image.In order to solve this problem,this paper uses the dynamic gradient feature of the reference graph to recover the area masked by smoke in the real-time image,and then find the position of the target by similarity measure.By comparing the results of this algorithm with the classical algorithm,the accuracy of the algorithm is verified.Fifthly,aiming at the characteristics of haze and smoke image,the image restoration algorithm is integrated into the target tracking algorithm to improve the accuracy of the tracking algorithm in the interference environment.At the end of this paper,the experimental results of this algorithm and the classical algorithm are compared,and the validity of the algorithm is verified.