【作者】 许波；

【导师】 李正明；

【作者基本信息】 江苏大学 ， 计算机应用技术， 2005， 硕士

【摘要】 随着实时成像跟踪技术在各个领域的广泛应用,特别是在军事和科学研究领域,对成像跟踪系统的要求越来越高,要求系统响应更快、精度更高。为了满足运动目标跟踪系统的高精度、实时性要求,本文以闻亭公司的WTC6201PA板为基础,建立了一个基于PC机和TMS320C6201的跟踪系统,以PC机作为主控计算机,TMS320C6201作为算法处理器的主从结构,充分利用DSP的强大数据处理能力对图像信息进行处理,实现实时跟踪。 在软件算法上,本文采用经典的ATR技术对目标进行检测识别。在图像分割中选取二维最大类间方差作为分割阈值度量函数,采用基于最优保存策略的自适应遗传算法进行算法的并行优化,取得了良好的分割效果。在对目标跟踪的过程中采用双模跟踪技术,根据目标的运动状态实时调整跟踪算法,同时针对模板匹配计算量大的问题提出一种先预测目标位置,再采用基于自适应遗传算法的快速匹配算法,有效的提高了运算速度。实验结果表明,采用上述算法和硬件组成的系统,能够较好的实现对运动目标的实时跟踪。更多还原

【Abstract】 With real-time image tracking technology have been extensively applied in various fields, especially in military affairs and science research, the demands for acquisition tracking and pointing systems are getting higher day by day. In order to satisfy the high accuracy and real-time requirement of the tracking systems of mobile objects, we implemented a tracking system with PC and TMS320C6201 based on WTC6201PA board of Wintech Technology Co.Ltd, in which, PC is used to be master computer and TMS320C6201 is used to be algorithm processor, the image is processed by fully using the powerful and high-speed data processing capability of DSP .So, a real-time tacking system can be realized.The classical pattern arithmetic ATR (Automatic Target Recognition) is applied in the tracking system. In the paper, a method of 2D maximum between-cluster variance (2DMVB) is used in image segmentation, at the same time, a adaptive genetic algorithm based on the elitist strategy is used to parallel-optimize the algorithm, the satisfied segmentation effects have been obtained. In the process of tracking, a dual-mode tracking algorithm is adopted. System can change the tracking algorithm automatically according to the moving state of the target. The template matching technique is confronted with large computational quantity, so we presents a new method, in which the location of the target was forecasted fist, then a adaptive genetic algorithm is introduced to do template matching, By using this algorithm, the matching time can be greatly reduced. The experimental result shows that the system we have designed with the algorithm and hardware of the above can track the moving target rapidly and accurately.更多还原