Research On Next Best View Based On Depth Image Information

Next best view problem has become one of main influence factors in the vision research fields such as target recognition, stereo matching, visual tracking and scene reconstruction, etc. Based on visual target depth image information, this paper comprehensively uses the theories and methods about computer vision, image processing, differential geometry, etc, and do studies on the next best view problem detection methods of visual target.Firstly, the concept of depth image, the methods of obtaining depth image, and several common principle methods of depth image pretreatment techniques are elaborated.Secondly, starting with the analysis of the depth image information and the surface mean curvature change, methods for triangular meshes are analyzed and studied on how to extract the feature points of depth image information, and designed a triangular meshes method based on surface mean curvature change and uniform sample. This method using triangles meshes information instead of depth image which extracts important features data for obtaining the next best view.Thirdly, this paper presents an improved method of establishing a triangular mesh data model based on the analysis of the existing STL(Stereo lithography) file and the characteristics of the triangular meshes information. The method makes numbers with the vertex and norm vector which avoids the shortcomings of the high duplication information and the operation inconvenience on the vertex about the existing STL file structure, establishing data storage models for features data extracted from the depth images.Finally, this paper proposed an algorithm of next best view based on occluded points by analysis triangular mesh data model, and obtained the specific position and angle of the next view. And simulated in Matlab7.0 environments, analyzed combining with the analysis of algorithms and the experimental results.