Object segmentation of indoor scenes using perceptual organization on RGB-D images
Chaonan Wang, Yanbing Xue, Hua Zhang, Guangping Xu, Zan Gao · 2016
In this paper, we address the problem of object segmentation, which is important for further scene analysis and scene understanding. To improve the accuracy of object segmentation from the images with indoor scenes, we propose a new algorithm which combined perceptual organization method with color information and depth information. In the proposed algorithm, firstly, the gPb-ucm method is used for initial segmentation, and then, color information and depth information are both used for perceptual organization. Color information and depth information are combined to achieve a complementary effect, and on the basis of image foremost segmentation, the segmentation result is modified by perceptual organization. Experimental results demonstrate the proposed method can effectively improve the accuracy of object segmentation.