Range Image Segmentation Using the Numerical Description of the Mean Curvature Values
In this paper we present a new efficient edge detection algorithm for the extraction of linear features in both range and intensity image data. In the proposed algorithm the distinguished points, which will comprise the edges, depend on the spatial analysis of the numerical description of the mean curvature values. The work was motivated by the fact that the optimality of edge detectors for range images has not been considered in the literature, some algorithms are limited to synthetic range images and will totally fail in the presence of noise, others which have been tested in real range images are complicated with large numbers of parameters. As it will be demonstrated, the algorithm features computational efficiency, high accuracy in the localization of the edge points, easy implementation, and robustness against noise. The algorithm was initially developed for range image segmentation and has been extended to segment intensity images with some modifications. The generality and robustness of the algorithm is illustrated on complex scene images with different range sensors.
Laser Scanner Range Image Segmentation Mean Curvature values Crease-step edge Free form objects
Yahya Alshawabkeh Norbert Haala Dieter Fritsch
Hashemite University, Jordan Institute for Photogrammetry (ifp), Universitat Stuttgart, Germany
国际会议
北京
英文
5577-5582
2008-07-03（万方平台首次上网日期，不代表论文的发表时间）