Abstract

This paper proposes a new supervised semantic edge and gradient extraction approach,which allows the user to roughly scribble over the desired region to extract semantically-dominanr and coherent edges in ir. Our ap- proach first extracts low-level edgelets(small edge clusters) from the input image as primitives and build a graph upon them,by jointly considering both the geometric and appear- ance compatibility of edgelets.Given the characteristics of the graph,it cannot be effectively optimized by commonly- used energy minimization tools such as graph cuts.We thus propose an efficient linear algorithm for precise graph op- timization,by taking advantage of the special strucrure of thte grapht.Objective evaluations show that the proposed method significanily ourperforms previous semantic edge derection algorithms.Finally,we demonstrate the effective- ness of the system in various image editing tasks