Iterative transductive learning for automatic image segmentation and matting with RGB-D data

Bei He, Guijin Wang, Cha Zhang

Abstract: Highlights•Iterative depth refinement and bi-layer classification is designed for segmentation.•We model the matting technique as an iterative transductive learning problem.•Coefficients between constraints and objective functions are adjusted adaptively.•A novel way to form the matting Laplacian matrix is proposed.