Graph-Based Local Kernel Regression for Image Editing

Graph Laplacian based framework is a powerful image modeling technology. In this paper, we present a new graph-based method. The key idea is to perform manifold regularization on data graph via local kernel regression. The graph Laplacian nature of our method is theoretically illustrated. Furthermore, based on the proposed framework, we apply our method to several image editing tasks, including remote sensing image fusion, interactive image segmentation and matting. Comparative experiments are conducted to validate the effectiveness and efficiency of our method.