【作者】 朱雯兰；

【导师】 隋成华；

【作者基本信息】 浙江工业大学 ， 通信与信息系统， 2005， 硕士

【摘要】 实时运动目标检测跟踪技术作为一门新兴的学科正在受到越来越广泛的关注。本文介绍了一个在实验室环境下运动小车跟踪系统的硬件设计与软件算法实现过程。该系统的核心是FPAG和DSP两部分,其中FPGA实现图像的预处理,处理后的图像送入DSP,通过设计的软件算法的运行,实现目标的检测与跟踪。 本文首先分析了引起图像模糊的因素,并给出了相应的处理对策。介绍了FPGA中的多种快速中值滤波算法。接着文章介绍了互相关,实时相减,自适应运动检测,图像分割等目标检测方法,分析比较了它们各自的原理和优缺点,并提出了一种改进的目标检测方法。本文还介绍了基于TMS320C6201 DSP的小车跟踪系统硬件平台的设计。最后文章给出了目标检测跟踪软件算法实现的详细分析。 本论文的创新之处在于提出了一种新型的,目的在于缩短目标检测时间,提高目标识别准确性的目标检测与跟踪算法。并通过VisualC++平台编程检验其准确性和实用性。更多还原

【Abstract】 As a new science, real-time detecting and tracking technology on moving object is being paid more and more attention. This paper introduces hardware design and software algorithm of a dolly tracking system in laboratory environment. FPGA and DSP are the two core parts of this system. Image preprocess is finished in FPGA, and then the processing result is carried into DSP. Object detecting and tracking is realized after running the software algorithm designed.First, this paper analyzes factors that cause image blurring, and presents corresponding disposing methods. Some fast median filter algorithms are advanced. Then this paper presents some object detecting methods such as co-correlation, real-time subtract, self-adapting method, image segmentation, etc. After analyzing and comparing their own principles, merits and defects, we advanced a new mended detecting method. Moreover, this paper introduces hardware platform design of dolly tracking system based on TMS320C6201 DSP. At last, we give更多还原