The Research On Infared Multi-target Detection And Tracking Technology

With the development of science of technology, especially the rapid development of infared technology. Infared video technology is widely used in guide of weapon system and IRSTS. Infared video technology has many advantages. Such as concealment, anti-jamming, sensitive detection and so forth. Infared multi-target detection and tracking has become important direction of the current research of infared search and tracking system.The detection and tracking of infared multi-target include three main procedures: image pretreatment, target detection, target tracking.In this paper,We study the multi-target detection and tracking in infared sequence images surveillance. Firstly, it will introduce the state and development of infared multi-target detection and tracking. we introduce the theories of infared video and analysis the characteristic of aerial infared target radiation. There are many methods of infared target detection are presented. then the performance of methods are analyze and validate through a group of experiments. We design a method of infared image target that suitable to complex background of air. the method of based on the morphologic background of gray scale has better performance. then the algorithms of target of labeling is introduced. Traditional multi-target tacking algorithms include: PDA(Probability Data Association)and MHT(Multiple Hypotheses Tracking).But the algorithms have many problems in application. This paper concentrate on tracking of multi-target in the air. It presents a algorithm of multi-target tracking base on dynamic estimation .This algorithm can track infared multi-target in the air。It establish a list of every target. And record the SNR, position. also forecast the possible target position in next frame. The program will search the target position in possible region and associate the target data, track multi-target.This paper also studies the complex problem such as two target overlap, target disappears and new target appears. Experimental results show that the performance of multi-target detection and tracking, algorithm works well in tracking infrared multi-target.