Infrared Target Detection And Tracking Algorithm

Infrared imaging guidance system, with strong anti-interference capability, working day and night, good advantages of subtle, strong viability, is an important research direction in the field of infrared image and video processing, using pattern recognition, image fusion processing, automatic control and artificial intelligence and computer application technology. From twenty-one, it is widely used in many Aerospace and aviation, traffic monitoring and military guidance field. As the key technology of infrared imaging guidance system, many experts and scholars, for many years, put a lot of effort on the technology of infrared target detecting and tracking, and, put forward a lot of infrared target detection and tracking algorithm.Infrared target detection and tracking is to find out target position, velocity, acceleration and other parameters in every frame of the target infrared video sequence. In practical application, infrared target detection and tracking should not only be affected by background and noise but also be affected by infrared imaging system noise introduced its own system, and, because of remote shooting, the target in the infrared images of video sequence is low SNR, no enough target shape, color and other information can be to use. The above problems are difficulties of infrared target detection and tracking, and, unsolved key.Nowadays, many domestic and foreign scholars devotes to the research on infrared target detection and tracking algorithm, put forward a number of Innovative and directive algorithm, such as Kalman, Meanshift algorithm, genetic algorithm, which are achieved good results. In this paper, on the basis of the previous research on infrared target detection and tracking algorithm, the main research work and innovations are as follows:(1) A novel target tracking algorithm based on Kalman prediction and top-hat transformation is proposed for the detection and tracking small target in infrared video sequences with low SNR and complex background. Kalman prediction is applied to predict the target position in the current frame. Top-hat algorithm is used to search the optimum target position in the suitable prediction area of current frame, which reduces the tracking computational time. The proposed method has the robust ability to track the moving object in the consecutive frames under some kinds of real-world complex situations such as the moving object disappearing totally or partially due to occlusion by other ones, fast moving object, changing the direction and orientation of the moving object, and changing the velocity of moving object suddenly. The proposed method is an effective video object tracking algorithm.(2) On the basis of analyzing the characteristics of infrared target under complex background, puts forward a kind of improved tophat algorithm for target detection. The algorithm first uses the edge characteristics of different structure of erosion and dilation algorithm of infrared small target detection, and then uses the particle filtering with genetic algorithm for small target tracking. The proposed algorithm has the advantages. It can better remove large area background and noise from the single frame infrared images, keep dot small target, by improved tophat method; particle filter algorithm is not affected by the linear model and the Gauss hypothesis constraints, so the infrared object tracking in video sequences have a certain stability, and, on the particle filter resampling the genetic algorithm not only guarantees the effectiveness of particle but also the diversity of particles, which improves the particle filter for infrared target tracking accuracy.(3) In the study of particle degeneracy problem deeply and learning the algorithm of Meashift, we add the Meanshift thought into particle filter algorithm. According to the weight of the particle, each particle is packet processed. Choosing the group Center for Meanshift iterative avoids that all particles are involved in Meanshift iteration, which reduces the computation complexity. According to tracking of each frame, the proposed method adjusts adaptive sampling number and sampling range, which improves the tracking accuracy. Experiments show that the method has good tracking accuracy, and has certain application value.