Research On Infrared Small Target Detection Algorithm In Sky Background

Compared with optical system, the environmental suitability of infrared imaging system is superior to visible light systems, especially the ability to work under bad weather during the night. Infra-red waves is longer than radar, which will guarantee the acquisition of target images with high resolution. Therefore, it plays an increasingly important role to capture and track the radiation energy of target itself to achieve precise guidance technology. And it is necessary that the Infrared imaging guidance system to be armed with efficient and accurate infrared small target detection device, which is very important to detect and track the suspicious target. Therefore, the small target detection in infrared image technology has always been one of the hot topics in target detection.The infrared small target detection is the detection of the small target which is the isolated points or spots occupying just several to a dozen pixels in the field of view and shows no shape features, amplitude distribution etc. due to the long distance of imaging system and target. Because of the characteristics of the small target itself, its detection and tracking is much difficult than big targets. Therefore, some introductory and innovative infrared small target detection algorithms have been proposed by a lot of experts and scholars.Based on the research of predecessors, this paper divides the existing detect before track (DBT) into the theory of machine learning technology based methods and image filtering based methods and proposed 3 effective infrared small target detection algorithms after researching the image filtering based methods:(1) A median filtering and gradient sharpening based method has been proposed by the fact that the background and target with noise occupy the Low-frequency component and the High-frequency component respectively. Firstly, part of the clutter noise in the original infrared image is filtered by median filter to get the background component by image smoothing. Then subtraction is applied between the filtered image and the original image (Background subtraction) according to the strong correlation between the background pixel gray value and the pixel gray value in its neighbourhood and the poor correlation between the target pixel gray value and the pixel gray value in its neighbourhood. At last the residual image is processed by Gradient sharpening method to detect target with threshold value limitation. The experimental results show that our method can not only remove background clutter effectively but also make the targets prominent with the targets detected.(2) The original infrared image is considered to be the degradation of the image after sharpening processing. Image restoration techniques is innovatively applied in the image smooth filtering and denoising process, which is to build degradation model of the unsharp filter (Anti-Laplace filter). And the inverse filtering (constrained least square filtering) filtering approach is used as recovering filter to reduce noise and improve image signal-to-noise ratio and image contrast. At last the background suppression and target enhancement technique proposed in Section 2 is used to suppress the background clutter detection and detect the infrared small target. The experimental results demonstrate that the proposed method can conquer the interference of clouds and exhibit robust properties with no track loss and the interference of false alarm points.(3) An infrared small target detection algorithm based of mathematical morphology and pipeline filter technology is proposed. The Tophat transformation of mathematical morphology is applied to locate some candidate target points by the gray value of the target pixel. Then the prediction location of target is obtained by the trajectory consistency and continuity of infrared small target motion to implement the infrared small target detection and tracking. The experimental results demonstrate that the proposed method is capable to achieve a better performance in infrared small target detection and tracking at high signal-to-noise-ratio.