Research On The Detection And Tracking Of Infrared Small Dim Targets

With the improvement of the status of air supremacy in modern warfare, the development of both monitoring alarm and infrared guidance attracts universal attention by the national military-industrial sector. Due to the limitations of the early radar monitoring alarm system, an infrared monitoring alarm system based on the principle of infrared thermal imaging has been widely applied. The system which combines technique photoelectric sensing, control and image processing has advantages of passive receiver, work thought day and night, strongly anti-jamming capability and so on. In order to provide adequate reaction time, targets must be discovered and validated as soon as possible. So detecting and tracking of small dim targets has become the core technology of infrared monitoring alarm systems.The paper aims at the detecting and tracking of infrared small dim targets in sky. First of all, the paper analyses the characteristics of the cloud background and come up with a method based on the complexity circular template to deal with the detecting of small dim targets. The principle of using information entropy to describe the complexity is introduced. The information entropy weighted by variance is used to suppress the region inside of clouds. To enhance the inhibitory effects of the cloud edge, a method based on the complexity circular template is put forward. Therefore targets can be detected effectively.Secondly, due to imaging characteristics of small dim targets, a method based on multiple feature fusion has been come up with. The paper introduce the theory of feature fusion and the target features used in feature fusion. Local contrast mean difference, local average gradient strength and local gray discreteness is used to describe targets, then they will be fused to highlight the targets’ feature, so targets can stand out from the background.Thirdly, according to ideas of global filtering, a method based on Singular Value Decomposition has been put forward to detect small dim targets. The paper introduces the algorithm of singular value decomposition and the method to make it into reality, then analyses the properties of singular value between targets and background, and comes up with a selection method of effective singular value to Guide the image reconstruction. So we can obtain the information of targets with little background and noise.Finally, a small dim targets tracking method which is a combination of a pipeline filtering and a gate tracking is proposed. The paper introduces the theory of a pipeline filtering and the way to determine the trajectory of suspected targets in order to exclude false alarms. Then it makes use of tracking gate in order to exclude irrelevant information and noise outside the gate to track targets. Since the combination effects of a pipeline filtering and a gate tracking, targets can be tracked availably even if targets may be disappeared for short time.