【作者】 翟鸣；

【导师】 梁栋；

【作者基本信息】 安徽大学 ， 信号与信息处理， 2005， 硕士

【摘要】 三维数据测量技术是计算机视觉领域的重要课题,在虚拟现实、文物保护、机械加工、影视特技制作、计算机仿真、服装设计等领域有着广泛的应用。基于结构光的三维数据测量方法是目前三维数据测量研究领域中的热点研究方向,该方法在进行测量时不需要与待测物体相接触并且测量速度快、成本低、精度高,具有极大的社会需求和发展前景,因此基于结构光的三维数据测量方法的研究不但具有重要的理论意义而且具有广泛的应用前景。 本文分析了当前三维数据测量技术的现状,对基于结构光的三维数据测量方法进行了较为系统的研究,主要包括测量系统的测量原理、测量系统的标定、光栅检测与索引、三维数据融合等。本文所完成的具体研究成果如下: 1.提出了一种基于平行结构光的三维数据测量方法,该方法不需要标定摄像机矩阵,使用平面单应原理求取待测物体表面的三维坐标。 2.对测量系统的测量原理和标定方法进行了研究,在文献[40]中整体标定方法的基础上,针对测量过程中摄像机参数改变时系统的重新标定问题,从理论上探讨了一种解决方法,可以一定程度上简化测量系统的重新标定过程。 3.对光栅的检测和索引方法进行了研究,主要包括:①对彩色光栅的检测和索引方法进行研究,使用Canny算子提取光栅边缘,利用光栅边缘的梯度信息来建立光栅索引;②借鉴彩色光栅索引方法的原理,提出了一种基于黑白光栅的检测和索引方法,依据物体在光栅照射前后灰度值的不同提取光栅条,使用关系矩阵描述光栅条间的相邻关系建立光栅索引,一定程度上降低了对测量环境和待测物体颜色的要求;③对Subdivision pattern光栅检测和索引方法进行研究,每次只提取不发生混叠的光栅边缘,一定程度上减少了光栅的误配。 4.对光栅边缘的精确定位和噪声消除问题进行了研究,采用Canny算子来精确定位光栅边缘,采用区域生长方法消除噪声,一定程度上解决了光栅边缘的精确定位和噪声消除问题。 5.对旋转平台三维数据融合方法进行了研究,针对其中的关键问题,即旋转平台旋转角度的计算问题,提出了一种基于图像消影点的旋转角度计算方法,并从理论上探讨了一种基于相关匹配的旋转角度计算方法。更多还原

【Abstract】 3D data measurement which is an important subject in the research of Computer Vision is wide used in many areas such as VR (virtual reality), antique protection, film industry, manufacture industry, CAD, CAM and clothing industry etc. Among the methods of 3D data measurement, the method based on structured light is a popular and promising subject, and it is in great need at present because it can get accurate 3D data from an object cheaply without contacting it. Consequently, it is significant in theory and practice to study on this method.The status of 3D data measurement at present has been analyzed detailedly and the method of 3D data measurement based on structured light has been studied systemicly in this paper, including the measurement principle, system calibration, image raster extracting and indexing, 3D data integration etc. The main work of thispaper is as follows:1. The method of 3D data measurement based on parallel structured light ispresented in this paper, in which the homograph theory is used to compute 3D coordinates of an object without calibrating the camera.2. The measurement principle and the calibration of the measurement system arestudied in this paper. The recalibration of the measurement system based on document [40] is discussed in theory, in order to decrease the workload when the camera parameters are changed in measurement.3. Much work has been done on the research of image raster extracting andindexing: in the study on extracting and indexing of the color raster, Canny edge detector is used to extract raster edge and the gradient information is used to index; the method of extracting and indexing of the black and white raster is presented, which is based on the principle of extracting and indexing of the color raster. In this method the contrast between the images of the object under the white light and under the raster light is used to extract image raster, and the relation matrix which describes the relation of the raster is used to index; in the study on extracting and indexing of the subdivision pattern raster, only the unpolluted edge is extracted every time, in order to eliminate wrong index.4. In the study on edge locating of image raster, Canny edge detector is used to locate the raster edge accurately. In the study on noise elimination, the method of region-grow is used to eliminate noise. Experiments show our methods are effective.5. In the study on 3D data integration by turntable, one method based on vanishing point is presented, which can compute the rotation angle of the turntable from the image, and the other method based on correlation is discussed in theory.更多还原