基于深度图像利用遮挡信息确定下一最佳观测方位

Determining Next Best View Based on Occlusion Information in Depth Image

文中提出一种新颖的利用深度图像中的遮挡信息确定下一最佳观测方位的方法.该方法首先从某一观测方位获取视觉目标的一幅深度图像,然后根据已获得的深度图像中的遮挡信息确定出下一最佳观测方位.主要贡献在于:(1)提出深度图像中最大深度差相邻点的概念,利用其与深度图像中的遮挡边界点可获取遮挡区域外接表面信息;(2)一种基于投影降维思想的遮挡区域外接表面最佳小平面集合的确定方法,用于确定下一最佳观测方位;(3)一种基于最佳小平面集合的下一最佳观测方位确定算法.所提方法无需预先获取视觉目标的先验知识及将摄像机的观测位置限定在固定表面上,适用于具有不同型面的视觉目标.实验结果验证了所提方法的可行性和有效性.

This paper proposes a novel next best view approach based on occlusion information in a depth image.At first,a depth image of visual object is obtained from one view,and then the next best view is determined by the occlusion information in depth image.This work is distinguished by three contributions.The first contribution is the concept of maximal depth difference adjacent point,which is combined with the occlusion boundary point to obtain external surface information of the occluded region.The second contribution is a determining method based on projection dimension reduction for the optimal small plane set of the occluded region＇s external surface,which will be used in the process of determining the next best view.The third contribution is a specific next best view algorithm based on the optimal small plane set.The proposed approach does not need the priori knowledge of visual object or limit the observation position of the camera on a fixed surface.In addition,the approach is suitable for the visual object with different surface.Experimental results demonstrate its feasibility and effectiveness.