Research On Infrared Small Target Detection Algorithms And Simulation Verification Based On Mobie Pipe

The war comes along since the human civilization. With the continuous development of industrial technology, the form of war changes from face to face to remote attack(air attack). For a country, self-evident air defense is significantly importance; The key technology of the missile is the detecting and tracking of small infrared target. The technology of the detecting and tracking of small infrared target is an important breakthrough after radar because of its advantages of high quality of image, fast speed of reaction, strong resistance to electronic interference, good capacity of hiding, and compactness of structure."Week" and "small" are the two characteristics in the concept of small infrared target. "Week" means the week energy accepted by the detector. "small" not only means the small size of the target, but also means the small area on the imaging system. In recent years, a large amount of methods have been proposed in the area of infrared small target detection in domestic and overseas by experts and scholars, such as Kalman filter, mean shift filter, information entropy filter, Gaussian kernel regression filter which have achieved good effect in practical application. Due to the algorithm of principal curvature filter just proposed by Zhao Yao is superior to other algorithms on the overall performance, the paper puts forward a new kind algorithm combining the principal curvature filter and mobile pipe. Through numerical simulation and hardware simulation, this algorithm improves the performance than the principal curvature filtering algorithm on the three kinds of performance index of time, probability of detection, probability of false. The principal content of this paper is listed below:(1) First the paper explains the algorithm of principal curvature filtering in detail and get some defects. Based on these defects, and the target’s characteristics of the continuous motion and consistent trajectory, the paper puts forward the algorithm combining the principal curvature filter and ordinary pipe. This algorithm’s main idea is that choose the previous frame’s target as the center to create the pipe, then detect the next frame’s target in the pipe area which has great improvement on time performance for the algorithm. In addition, the algorithm has the function of distance judge which can exclude the some interference of false points, reduce the probability of false points.(2) The paper proposes a new kind of algorithm combining the principal curvature filter and mobile pipe. This algorithm makes better use of the continuous motion and consistent trajectory, at the same time, adopt the characteristic of hessian matrix’s maximum and second maximum to exclude the interference of fixed noise. The algorithm can detect a real pipe through the some or dozens of frames of images’ detection after setting pipe’s length. Also the algorithm can predict the real target’s position through the algorithm of line fitting when the target is blocked or disappear in short period of time. The algorithm improves the performance than algorithm combining the principal curvature filter and ordinary pipe on the three kinds of performance index of time, probability of detection, probability of false through the ROC curve.(3) In this paper, the Algorithm is not only be verified through the simulation software such as matlab, but also be verified through the hardware. First use the camera to acquire the image of small infrared target and convey the image to the hardware. Then the hardware will get the position of target point through the algorithm combining the principal curvature filter and mobile pipe. At last the soft will detect the point with the black mark and compare the real point.