Recognition of transparent objects using 3D depth camera

In this paper, we propose a method for recognizing transparent objects by depth data obtained by using `Intel Realsense R200'. When a transparent object is measured with Relsense, a black region in which there is no depth data due to the characteristics of the transparent object is generated. By using morphology operation, we remove the unnecessary depth data of the region and proceed segmentation. After segmentation, edge detection is performed. Using the resulting edge data, we determine whether the object is a transparent object or an opaque object by judging the similarity of the color composition of the edge data foreground and the adjacent background.