【作者】 李凯；

【导师】 张丕状；

【作者基本信息】 中北大学 ， 信号与信息处理， 2008， 硕士

【摘要】 计算机视觉技术通过分析空间目标的多幅图像得到目标的空间信息,其目标是实现三维场景的感知、识别和理解。现有的计算机视觉技术主要采用复杂的相机标定技术实现,测量精度取决于相机标定的精度,测量精度不易控制,在实际应用中操作复杂等缺点。本论文借鉴计算机视觉技术的特点建立了适用于空间静态目标测量和动态目标测量的测量系统,利用图像处理技术实现了空间目标测量的自动化。本论文建立的测量系统不需要复杂的相机标定,同时具有对测量场地要求低、实现简单、精度高、稳定性高、应用范围广的特点。本文首先分析了双目立体视觉技术,分别建立了双相机光轴面垂直测量系统和优化双相机测量系统,同时为测量系统建立了完整的数学模型,两种测量系统不需要在测量时对相机参数进行标定,对相机放置位置要求较低,适用于测量环境不确定的目标测量;其次,分析了相机标定方法和相机标定时标定点的设置方法,根据标定标杆和标定模版的特点建立了标定模块,它具有标定点空间相对位置稳定,且满足现有标定算法的要求;然后,分析了图像特征点检测技术和特征匹配技术;最后本论文通过静态建筑物中目标点的空间位置测量验证了双相机光轴面垂直测量系统的正确性,通过静态建筑物中目标点的空间位置测量和空间多目标运动速度和轨迹的测量验证了优化双相机测量系统的正确性,通过实验结果证明了本论文提出的测量系统满足实现空间多目标运动参数的测量要求。更多还原

【Abstract】 Computer vision is the method to acquire the 3D space information of objects by multiple images. The existing computer vision technology mainly uses the complex camera calibration to realize survey, the measuring accuracy is decided by the precision of camera calibration, and not easy to control, operates complex in the practical application.The dissertation use computer vision technology characteristic to establish two kinds of novel simplified measurement systems, which is suitable to survey spatial static goal and dynamic goal, has realized the automatic measure by image processing technology. The measurement systems of this dissertation do not need the complex camera calibration, and have characteristic of low requirements for measuring position, simple, high precision, high stability, wide application scope.First,the binocular stereo vision measurement systems are studied in dissertation, two kinds of novel simplified dual-camera measurement systems, namely one system in which dual-camera measurement system of vertical camera axes’surface and the other in which excellent dual-camera measurement system, is proposed and the corresponding mathematic model is established. Second, the camera calibration methods and the establishment method of calibration points are studied in detail. Second, a calibration module is proposed, base on the characteristic of calibration pole and calibration template, which has stable location among points, and is suitable to the requirements of existing calibration algorithm, at the same time it is simply in using.Third, image feature detection technology and feature matching technology is studied in dissertation.The last, The dissertation test the dual-camera measurement system which has vertical surface of cameras axes is right by measuring the space points coordinates of building; test excellent dual-camera measurement system is right by measuring the space points coordinates in building and measuring speed and path of space multi-target; test the measurement can satisfies of measurement requirement of parameter for space moving multi-target by the experimental results.更多还原