空间翻滚非合作目标位姿及运动参数视觉识别技术研究

发布时间：2018-06-09 15:17

本文选题：非合作目标 + 双目立体视觉测量　；参考：《哈尔滨工业大学》2017年硕士论文

【摘要】：随着太空中不受控的空间碎片、失效卫星等空间废弃物越来越多,以清理空间碎片或修复失效卫星为目的的在轨服务技术得到了广泛关注。而作为在轨服务中的关键环节,对目标位姿及运动参数测量技术的研究必不可少,本文对基于双目立体视觉的目标位姿及运动参数识别技术进行了研究。首先,在深入探讨了双目立体视觉测量系统相关理论的基础上,建立了摄像机成像模型,并以此为基础构建了双目立体视觉测量模型。然后,研究了双目立体视觉测量过程中图像处理的关键技术——特征提取和立体匹配算法。经过对Harris角点检测算法的研究,针对其无法精确检测所需角点和阈值设定困难的缺陷,结合本课题目标特点,本文提出了一种改进的Harris角点检测算法。此算法首先利用Hough变换检测出的直线信息计算疑似角点区域信息,并以此区域信息分割图像,在分割的图像中进行角点判定,这不仅提高了检测精度,同时降低了算法对光照的敏感性。为提高立体匹配的正确率,本文提出一种将直线特征与SAD匹配算法相结合的立体匹配方法。在完成图像处理相关技术的研究后,文中建立了位姿解算模型,并在此模型基础上,推导了运动参数解算模型。最后,设计了目标识别实验系统,包括非合作目标模型的制作,非合作目标运动平台以及视觉测量系统的搭建等。完成了测量实验,对文中提出的位姿解算模型与运动参数模型的效果进行验证,实验结果表明:本文搭建的双目立体视觉测量系统精度较高,稳定性好。
[Abstract]:With the uncontrolled space debris in space, more and more space debris, such as invalid satellites, have been paid more and more attention. The in-orbit service technology, which aims to clean up space debris or repair failed satellites, has been paid more and more attention. As the key link of in-orbit service, it is necessary to study the measurement technology of target pose and motion parameter. In this paper, the recognition technology of target pose and motion parameter based on binocular stereo vision is studied. Firstly, based on the theory of binocular stereo vision measurement system, the camera imaging model is established, and the binocular stereo vision measurement model is constructed. Then, the feature extraction and stereo matching algorithms, which are the key techniques of image processing in binocular stereo vision measurement, are studied. Based on the research of Harris corner detection algorithm and the defect that it can not accurately detect the required corner and threshold, an improved Harris corner detection algorithm is proposed in this paper. The algorithm firstly uses the line information detected by Hough transform to calculate the suspected corner area information, and then segments the image with the region information, and determines the corner point in the segmented image, which not only improves the detection accuracy, but also improves the detection accuracy. At the same time, the sensitivity of the algorithm to illumination is reduced. In order to improve the accuracy of stereo matching, this paper proposes a stereo matching method which combines straight line features with SAD matching algorithm. After the research of image processing related technology, the pose and pose solution model is established, and the kinematic parameter model is derived based on the model. Finally, the experimental system of target recognition is designed, including the making of non-cooperative target model, the construction of non-cooperative target motion platform and vision measurement system. The experimental results show that the binocular stereo vision measurement system presented in this paper has high precision and good stability.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】