Range Image Segmentation Based On Curvature And Normal Vector

With the development of information technology, people pay more and more attention to the range image of the 3D information. Then processing results of range image further to people living in all aspects. Give people more convenience and huge developments. Image segmentation is one of the main research directions of image processing technology, it is precondition in image processing, in the other word, image segmentation will have a direct impact on the image processing results and final application, range image is one of important segment in range image study.Binocular measurement system is used for get the range image based on range image acquisition research in this paper. The method of improving median value filter is used for reduce image noise. Experimental results show that this method not only effectively remove the noise but also keep edge information features.According to the geometrical characteristics of range image, new way of normal vector of range image segmentation been taked. This method is different from the former methods .It don't need first to fitting of surface calculating the normal vector ,but use gradient eigenvalues of one-to-one normal vector for feature extraction undertakes breaking up. This method can avoid noise of fitting of surface era on the segmentation effect. Experimental results show that although very good reduces the noise influence but at the edges appeared deformation.In order to improve the edge detection quality, based on curvature and normal vector of range image segmentation been taked. It is primarily with range image geometric quantity (normal vectors, curvature) for feature extraction. Firstly According to curvature changes detect and mark image edges, secondly make gradient segmentation, finally fusion processing to get range image segmentation result. Through the experiment and evaluated system show this method which reduced the noise influence and have accurate edge, obtained best range image edges.