Infrared Weak Small Target Tracking Algorithm

From the twentieth century, infrared imaging guidance technology is widely used in military fields. It has the advantage of anti-jamming, concealment and high precision tracking performance. Infrared target detection and tracking as the imaging guidance system core technology has also been widespread concern. In practice, infrared target detection and tracking process becomes very complicated due to the impact of background and noise. When the imaging distance of targets are always far, the target occupy a few points or several pixel in the image and the signal is weak. Target submerged in the background and noise can not be detected. Therefore, target tracking problem has become the main research topic based on the complex background of infrared small.Nowadays, many scholars committed to infrared target detection and tracking of research work. They made a number of innovative and guiding target detection and tracking algorithms, such as kalman filtering algorithm; mean-shift algorithm; genetic algorithm and time domain filtering algorithms. These algorithms have good results in the infrared target detection and tracking. This dissertation deeply analyze the characteristics of the infrared small target, the background and noise, and then proposed three improved target detection and tracking algorithm based on complex background. The proposed algorithms of this dissertation are as follows:(1) A temporal and spatial filter algorithm is presented to detect Infrared weak point targets moving slowly in the scenarios with cloud clutter and noise. Firstly, IR image sequence is processed by the normalization preprocess. Secondly, a filter is used to process the image sequence in the temporal domain, and then a target enhancement algorithm is constructed to suppress the residual dots which are attributed to the cloud clutter and noise. Finally, the computer simulations are made to verify the performance of the presented algorithm, and the results confirmed its effectiveness.(2) An algorithm based on kalman filter and intensity variation function is presented for infrared target tracking. The algorithm overcomes the defects of target tracking based on low contrast. Firstly, the local maximum is extracted in the previous frame. Secondly, the kalman prediction method is used to predict the target point which is the center to create a subframe in the current frame. Thirdly, use the intensity variation function to track the target. Finally, the experiment verified the effectiveness and feasibility of the proposed algorithm.(3) An algorithm based on a kalman filter and genetic algorithm is presented for infrared target tracking. Although genetic algorithms for target tracking algorithm can effectively reduce the amount of computation, but also reduces accuracy and stability, especially in the case of complex background using genetic algorithm to track the target point, the target is easy to the lost. To solve this problem, a local maximum is firstly selected in the previous frame. Secondly, the kalman prediction method is used to predict the target point which is the center to create a subframe in the current frame. Thirdly, the initial population is established in the subframe. Local maximum and initial population are substituted in the genetic algorithm to track target. Finally, the experiment verified the effectiveness and feasibility of the proposed algorithm.