Infrared Small Target Detection Method

In recent years, with the development of the computer and the digital imageprocessing, the infrared images have wide application in the fields of the guidance,thetrack,the automatic control and the artificial intelligence. The study of the lowsignal-to-noise ratio and infrared small target's real-time detection is significant to themodern high-tech war. It can expand the coverage of sensors, detect target earlier,lock the target, and enhance the intellectualization of infrared image system as muchas possible. The difficulties of small target detection lie in: low signal-to-noise ratio,non-texture characteristic, and little reliable target information. The method of thedetection-before-tracks (DBT) by single frame image is difficulty to detect the smalltarget, but the tracks-before-detection (TBD) can detects the small target by the three--dimensional information of sequential images.This article studies the algorithm of tracked-before-detection based on thecharacteristic of infrared images. Infrared image preprocessing is important in theinfrared small target detection; it can suppress each image's background noise,enhance signal-to-noise ratio and reduce calculation. This article introduces severalpreprocessing methods and analyzes the advantages and a disadvantage by theexperiments. This article introduces the dynamic programming algorithm in detail,and proposes an improved algorithm that combines the multistage threshold ofmultiple hypothesis testing. The improved algorithm holds the accuracy of the DPalgorithm, at the same time, reduces the computation of system. It will have thepractical application in the detection of infrared small target in the future.