【作者】 闾海庆；

【导师】 邹峥嵘；

【作者基本信息】 中南大学 ， 资源环境遥感， 2006， 硕士

【摘要】 随着近景摄影测量技术的进一步发展,以及普通数码相机的分辩率不断提高,数码相机已经成为开展近景摄影测量中重要的影像数据获取设备。本文基于传统的近景摄影测量领域的一些研究成果,对普通数码相机的检校和影像数据处理问题进行了进一步研究,主要的研究成果为:1.利用室内控制场对数码相机进行量测化改造,利用三维控制场确定相机的畸变系数和内方位元素,利用二维控制场建立对应焦距的数字畸变模型。最后,对量测化精度进行评价。实验结果表明,采用本文的量测化方法精度能够满足高等精度要求的数字摄影测量项目。2.为了解决空间后方交会和相对定向中,由旋转矩阵中含有三角函数及角量造成迭代计算不收敛的情况,本文将Pope-Hinsken算法引入到相对定向中,用四个代数参数代替通常用的角量,从而对于任意角度拍摄的非量测数码影像在空间后方交会和相对定向中总能保证以较快的速度收敛。3.在处理多基线非量测数码影像数据时,本文是采用相对定向—模型连接—绝对定向的思路进行的,然后将定向结果作为光束法平差的初始值,进行数据处理得到所需监测点的点位信息。本文首次尝试将航空摄影测量中空中三角测量技术引入到近景摄影测量当中,通过多次实验结果的表明,在控制点的数量和分布合理的情况下,光束法平差的精度可以达到1/5000。4.多基线数码影像中控制点数量有限,必须通过模型连接将所有影像纳入到统一坐标系中来,才能进行绝对定向。这里模型连接是通过公共连接点在前后两个模型中的高程相等这个条件完成的,将连接后的整体模型进行绝对定向,最后将所有监测点坐标转到大地测量坐标系中。5.由于多基线光束法平差与航空摄影测量中的光束法平差理论相似,本文通过将控制点作为真值的方法进行光束法平差,并对控制点的数量和分布对平差结果的影像进行来详细的讨论。6.最后通过部分实例来验证多基线普通数码影像数据处理技术在实际应用中的精度。更多还原

【Abstract】 With the development of close-range photogrammetry and the increase of resolution, common non-metric digital cameras have been the important equipments for close-range photogrammetry. Based on original critical research results, we did some research on camera calibration and data processing. This paper researched on digital close-range photogrammetry based on multi-baseline normal digital images. And it introduced the following results.1. In order to know lens distortion of digital cameras, we took full use of control field. Research and experiments illustrated that the measurement and alteration of common digital camera is successful. This photogrammetric techniques and methods based on ordinary digital images can be applied extensively in engineering which requests a moderate accuracy.2. During the process of data processing, we take three steps which include relative orientation, absolute orientation and model collection. It will provide initial value for bundle adjustment. The results of the experiment give us a good suggestion. The precision of bundle adjustment could reach 1/5000 if the number of control-points is enough and keep a good distribute.3. Rotation in relative orientation is consisted of angles which is the main reason of bad results. However, Pope-Hinsken could solve this problem. It replaces the angles with four parameters. It is a good way for the relative orientation.4. Because the number of control Points in multi-base line images is limited, we should take the action of model collection in order to make all images in the same control conditions. Model collection is depend on the points’ elevation. After the conjunction of whole model, the next step is absolute orientation. Finally, we could convert the coordinate’s control conditions into geodetic system.5. The bundle adjustment in multi-base line images is similar to Aerial photogrammetric. We always consider the value of the control points to be real while we take the action of bundle adjustment. We also did some research on the relation between control points’ number and distribute.6. Finally, We take some examples to certificate the reliability of multi-base line images.更多还原