Research On Detection And Tracking Algorithm Of Infrared Small Target In Complex Urban Background

Infrared dim small target detection and tracking technology is one of the research hotspots in the field of information processing technology in recent years.It is widely used in the field of remote warning,infrared guidance,security monitoring and other fields.Compared with the sea sky background,the complexity of city background is greatly improved,more challenging and practical.In this paper,the detection and tracking algorithm of infrared small and weak targets in the infrared image sequences captured by the airborne dithering device in the complex urban background is studied.This paper is divided into four parts: infrared image analysis,preprocessing algorithm,detection algorithm and tracking algorithm.In the first part,the characteristics of infrared image are analyzed,including the principle of thermal infrared imaging and the analysis of crosstalk.Then,the mathematical model of infrared dim target is given and the complexity of urban background is discussed.In the second part,the algorithm of infrared dim small target image enhancement in urban background is studied,and a preprocessing algorithm based on crosstalk suppression is proposed.First of all,the electrical crosstalk degradation function model is established according to the characteristics of electrical crosstalk,and the classical LR deconvolution algorithm is used to solve the degradation model.Finally,the experimental results show that the preprocessing algorithm can effectively enhance the small and weak targets and improve the image definition.In the third part,the small and dim target detection algorithm based on the infrared image sequence is studied,and an improved algorithm for dim target detection based on ORB feature registration difference is proposed.First introduces the basic principle and main process of the original algorithm,and then from the two aspects of image registration and segmentation the improvement schemeis proposed.Combined with the Pyramid LK optical flow algorithm the image registration is improved and using the P-Tile method of object segmentation is improved.Finally,experiments are carried out to show that the improved algorithm can improve the registration accuracy,and has lower false alarm rate in complex urban background.In the fourth part,the algorithm of small target tracking in urban background is studied,and an improved algorithm based on KCF algorithm is proposed.Firstly,the principle of the KCF tracking algorithm is introduced,and the shortcomings of the algorithm are pointed out.Then the uniform linear infrared dim target motion model is established based on analysis,and then improved the KCF algorithm by fusion Calman filter,mainly to solve the occlusion or interference problems in the background of city.The experimental results show that the algorithm can effectively solve these problems.