Target Recognition And Tracking Of New Technologies And Models

The optoeleceronic image tracking system's tracking precision is excellent, and it has applied in many kinds of guidance weapon. The new-style antiaircraft weapon needs the support of the optoeleceronic image tracking system especially. Target recognition and tracking in the complex background are the main problems in this filed. On the demand of the real weapon task, in order to deal with targets in the complex background, this dissertation use the new theories in the field of image processing and computer vision to research the new technology of target recognition and tracking. The principle content is as follows:On the base of the energy-minimizing image segmentation, and by processing the image according to the row, we build a row energy function of images, and set forth the row energy-minimizing segmentation. With analyzing the area's movement, we can reject some clutter.Utilize principle component analysis (PCA) or singular value decomposition (SVD) to draw the image algebraic feature, and create the eigenspace, to classify and recognize the man-made targets in the complex background. Experiments declare this method is feasible and effective.Improve the traditional correlation matching method by robust statistics and affine transformation. Summarize the target tracking process on the base of mean robust difference (MRD) algorithm, and compare with MAD and Prod algorithm. The stability of this algorithm is good in many target-tracking experiments in the complex background.Introduce the hardware design of a real-time optoeleceronic tracking information processor. According to the process of the system and special demand of DSP software developing, we describe how the target-tracking algorithm brings into effect on the hardware platform. And design the big-target tracking module of system software to deal with the targets whose area is bigger than the limit in the complex background.