The Research Of Moving Targets Detection And Analysis In Infrared Image Sequences With Complex Background

With the development of infrared imaging techniques, infrared imaging is more and more used in object detection, recognition and so on, including recognition and tracking of moving targets. Detection and analysis of moving objects in image sequences are the processes that separate moving objects from still background and analyze them using the spatial and temporal relativities of serial images.Optical Flow computing don't require the rigorous corresponding relationship among features of sequential images, so this approach is widely used in computer vision field including detection and dynamic analysis of moving objects. Generally infrared images suffer from relatively high stochastic noise and non-uniformity clutter, yet the traditional optical flow technique is based on image intensity data, so the optical flow technique is seldom applied to detection and analysis of objects in infrared images. Making full use of the power of combining infrared imaging with optical flow model, we propose a moving object pre-detection algorithm based on supervised learning, image pair difference significance test and minimum cost Bayes rule. In this algorithm according to minimum cost Bayes rule we acquire a threshold of moving object segmentation through human-participant learning and apply this threshold to moving object detection in posterior images. On the basic of moving region detection, We also apply region labeling and local segmentation to those regions. Thus, the moving object detection is achieved and we get satisfying results. If combined with optical flow computing, this algorithm can highly reduce the computation amounts and enhance the computing rate.Based on moving object detection, we also make some researches on dynamic object analysis and propose a dynamic infrared imagery analysis method based on knowledge representation and scene modeling. This algorithm can efficiently apply global modeling of sequential images to tracking, pattern discrimination and behavior analysis of moving objects in infrared imagery sequences. Finally, we summarize the matters to be further studied.