Study On Image Segmentation And Moving Target Tracking

Research of digital image processing and computer vision is developing rapidly these days, and is widely used in civil and military applications. Automatic target detection and tracking is an active topic in this area. This thesis focuses on the key techniques of target segmentation and tracking algorithms of image on the sea, which is used in TV control and guide of aerocraft.After the state of arts of image segmentation technique is introduced, the key techniques of classical approaches (threshold and edge detection method) are presented. Then the performance of these approaches are analyzed and validated through a group of experiments. In views of some disadvantages of these methods, a method of infrared image segmentation that is suitable to complex background on the sea is designed in this thesis. The experiments results show that this method has better performance.On the basis of image segmentation, this thesis mainly discusses the algorithm of automatic target tracking in sequence images on the sea. Firstly, in terms of the characteristics of image on the sea, such as weak contrast between target and background, borderline between sky and sea, spoondrift and so on, a method called barycenter tracking which is based on intra-frame information from the sequence images is proposed, in which the threshold method is the key algorithm. The experiments show that this method has great performance of single target tracking in sequence images on the sea. Considering the correlation between the frames of sequence, we research the tracking methods which take advantage of both the intra-frame and inter-frame information, including predicting tracking machine based on Kalman filtering algorithm and matching tracking machine based on Mean-Shift algorithm. Both the methods can realize the appointed target tracking in multi-targets scene. Through the experiments, we can get the conclusion that the latter method has better performance than the former one.At last, the tracking algorithm runs in a hardware system of automatic target tracking. This system uses TMS320C6711 DSP (Digital Signal Processor) as CPU, and uses FPGA (Field Programmable Gate Array) to control the system...