VDESIGN: Toward image segmentation and composition in cave using finger interactions

Xiaoming Nan, Ziyang Zhang, Ning Zhang, Fei Guo, Yifeng He, Ling Guan

Abstract: The Cave Automatic Virtual Environment (CAVE) system is a fully immersive virtual reality system, which can provide users with a realistic experience and a large freedom of interactions. In this paper, we propose vDesign, a CAVE-based virtual design environment using finger interactions. Specifically, we focus on the function of image segmentation and composition in the vDesign system. In vDesign, the user wears a marker on each hand. The interactions of the user are triggered based on the real-time positions of the markers. We design multiple finger interactions for image segmentation and image composition. In image segmentation, the user can use the right finger to select the interested object and the left finger to select the unrelated background. Based on the user's selection, a graph-cut based image segmentation is employed to extract the interested object from the image. In image composition, the user can move, rotate, and scale the segmented objects with fingers and combine them together into a final image. We implemented the vDesign prototype and conducted experiments to compare the finger interactions and the traditional wand interactions. The experimental results demonstrated that the proposed finger interactions can provide faster and more accurate interactions compared to the traditional wand interactions.