复杂背景下足球守门机器人目标追踪技术研究

发布时间：2018-08-22 15:33

【摘要】：随着时代的发展,基于机器视觉的目标追踪技术在人们的日常生活和工业生产中发挥着越来越重要的作用。但在实际目标追踪技术应用过程中,由于目标背景一般较为复杂并且可能存在动态干扰,导致算法的实时性和追踪精度相互制约。因此,研究复杂动态背景下的目标追踪算法具有重要的理论意义和现实意义。本文针对复杂动态背景下的目标追踪问题,采用足球守门机器人系统作为实验平台,对相关目标追踪技术和算法进行了研究工作。针对双目系统的实现原理,本文进行了详细的公式推导和介绍。对于光学镜头导致的图像畸变,本文总结了三种畸变模型,以便设计图像矫正方法。对于多种立体匹配的度量方法进行了详细的分析和比较,提出了基于LoG算子的过零匹配方法和基于图像结构张量的立体匹配算法。针对复杂背景下的目标追踪算法,本文提出了基于改进光流计算方法的目标追踪技术,可以基本实现系统实时性的要求。本文又提出了基于内嵌Mean Shift算法优化的粒子滤波算法,用于复杂背景下的快速目标追踪。针对粒子滤波算法中粒子有效性和多样性退化问题,本文提出了基于最大模糊熵高斯粒子聚类优化算法以及对于粒子的二次重采样方法,可以有效的解决粒子性能退化问题,并且兼顾了实时性的要求。最后针对文章中提出的相关研究理论和目标追踪算法,基于足球守门机器人系统进行了人机协同测试,对目标追踪算法进行了验证和检验,并针对对实验结果进行了统计与分析。
[Abstract]:With the development of the times, target tracking technology based on machine vision is playing a more and more important role in people's daily life and industrial production. However, in the application of target tracking technology, the real-time performance and tracking accuracy of the algorithm are restricted due to the complex background of the target and the possible dynamic interference. Therefore, it is of great theoretical and practical significance to study the target tracking algorithm in complex dynamic background. In order to solve the problem of target tracking in complex dynamic background, this paper uses the soccer gatekeeper robot system as the experimental platform to study the related target tracking techniques and algorithms. According to the realization principle of binocular system, this paper deduces and introduces the formula in detail. For the image distortion caused by optical lens, three kinds of distortion models are summarized in order to design image correction methods. The measurement methods of stereo matching are analyzed and compared in detail. The zero-crossing matching method based on LoG operator and the stereo matching algorithm based on image structure Zhang Liang are proposed. Aiming at the target tracking algorithm in complex background, this paper presents a target tracking technology based on improved optical flow calculation method, which can basically meet the real-time requirements of the system. In this paper, a particle filter algorithm based on embedded Mean Shift algorithm is proposed for fast target tracking in complex background. In order to solve the problem of particle efficiency and diversity degradation in particle filter, this paper proposes a Gao Si particle clustering optimization algorithm based on maximum fuzzy entropy and a quadratic resampling method for particles, which can effectively solve the problem of particle performance degradation. It also takes into account the requirement of real-time. Finally, in view of the related research theory and target tracking algorithm proposed in this paper, based on the soccer gatekeeper robot system, the man-machine cooperation test is carried out, and the target tracking algorithm is verified and tested. Statistics and analysis of the experimental results are carried out.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP242

【相似文献】