Occlusion Boundary Detection Based On Unsupervised Online Learning And Avoid Occlusion Based On Occlusion Information

In our daily life, Occlusion problem is everywhere. Today, rapid advances in video technology, Occlusion has become an major problem in the visual estimation, 3D reconstruction and pattern recognition, how to detect occlusion in image and how to avoid the occlusion. It became an important question in the image detection research, in order to solve these problems, this paper presents a novel method to detect occlusion boundary, a novel occlusion boundary detection approach based on unsupervised online learning and determining next best view based on occlusion information are proposed.Firstly, the occlusion related features of the frame to be dete cted in video sequences are extracted and the time length corresponding to the frame is calculated. the pixel points' occlusion related information in the frame to be detected is obtained by using Hedge Algorithm and combining time length and different occ lusion features. The occlusion related information of different features are voted to accomplish occlusion boundary detection of current frame.Secondly, the detection result of current frame is used to estimate the feature weight for next frame based on O nline Boosting idea to realize the detection of subsequent frames. The proposed method changes the feature weight through online learning idea to accomplish occlusion boundary detection, and does not need to obtain priori knowledge of video sequences in advance.Thirdly,When we detected image occlusion boundary, in order to avoid the occlusion, we can determine the next best view position based on existing information about the occlusion? Because the depth image more accessible than the color image to get information of three-dimensional scene, for the depth image, determining next best view based on occlusion information and plane modeling method are proposed..Finally, Experimental results show that, compared with existing method, the proposed method has higher accuracy and better generality.And, comparison and corresponding analysis for method of occlusion boundary detection and determining next best view are also presented in this paper.