A novel hand-drawn sketch descriptor based on the fusion of multiple features

Constructing a distinctive and robust sketch descriptor is one of the most challenging problems in sketch based applications. In this paper, a new hand-drawn sketch descriptor is proposed. The proposed descriptor utilizes the statistic information of multiple features and bag-of-features representation to achieve translation and scale invariance and rotation robustness. The proposed descriptor also encodes the information entropy distribution of information point located on the contour, which can describe the intrinsic property of sketch better and is more robust to noises. Experimental results demonstrate the validity and efficiency of the proposed sketch descriptor.