Infrared Small Target Detection

Infrared dim small target detection is the key technology of infrared early warning system. Because the distant target imaging area is very small and energy is weak, it is difficult to detect the target from a complex background. Thus, the infrared small target detection in low SNR complex background is an important technical problem for Military Applications. The infrared small target detection is researched in this paper, and some methods are proposed to solve this problem, details are as follows:First, it introduces different filtering algorithms and analyzes the performance of them. Then it improves the traditional filtering algorithm and proposes Guard-band filtering algorithm. Through comparative experiment, it proves the effectiveness of the algorithms. Then, it proves that Guard-band median filtering algorithm can suppress the background efficiency and stability.Next, This paper studys the infrared small target detection in two ways:single-frame detection and multi-frame detection. In single-frame detection, it introduces two algorithms: First, on the basis of image pre-processing and energy accumulation, it uses image segmentation algorithm to finish target detection; Second, a new method based on robust statistics is proposed, it transformes target detection problem into outlier identification problem. It completes target detection by identificating the outlier. This algorithm does not need to pre-process the infrared image, which expands the thinking of infrared dim target detection in a single frame image. In multi-frame detection, it takes advantage of target motion characteristics and proposes historical track algorithm. First, it enhances target energy by airspace and time domain. Then it calculates the target historical movement trajectories, and removes dead spot noise by noise mask image. Finally, it achieves the small target detection through connected domain detection.Finally, it studies target tracking algorithm based on PDAF. It predicts the target's motion state and updates the filter's parameter of small target based on Kalman filter, and analyzes the tracking performance by experiments.