期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2010
卷号：XXXVIII - Part 3B
页码：104-109
出版社：Copernicus Publications
摘要：In this paper, a new method for computing surface parameters, especially the surface roughness, is presented. This method is designed for easily reconstruct and extract informations from a collection of photos taken without any constraints. This absence of constraints is possible since camera calibration can be computed with bundle adjustment auto-calibration methods. 3D information can then be retrieved with triangulation techniques from the disparity maps computed for each image pair. This paper proposes a new statistically grounded extraction of the roughness directly on the 3D point cloud. Joining 3D and image processing methods, the roughness can be computed only on certain objects with image segmentation. The results are shown on different datasets proving the method robustness.
关键词：Stereo-vision; point cloud; soil roughness; bundle adjustment; epipolar geometry; auto-calibration
