An L1 Image Transform for Edge-Preserving Smoothing and Scene-Level Intrinsic Decomposition   Sai Bi               Xiaoguang Han            Yizhou Yu   The University of Hong Kong   ACM Transactions on Graphics (Proceedings of SIGGRAPH 2015) Figure 1. Our algorithm for piecewise image flattening facilitates both edge-preserving smoothing and intrinsic decomposition. Two examples of edge-preserving smoothing are shown in (a)-(d), and one example of intrinsic decomposition is shown in (e)-(h). Intrinsic decomposition enables image editing, such as re-texturing (i). Original images courtesy Flickr users 47765927@N06 (a), 132341054@N03 (c) and37213589@N08 (e). Abstract Identifying sparse salient structures from dense pixels is a longstanding problem in visual computing. Solutions to this problem can benefit both image manipulation and understanding. In this paper, we introduce an image transform based on the L1 norm for piecewise image flattening. This transform can effectively preserve and sharpen salient edges and contours while eliminating insignificant details, producing a nearly piecewise constant image with sparse structures. A variant of this image transform can perform edge-preserving smoothing more effectively than existing state-ofthe-art algorithms. We further present a new method for complex scene-level intrinsic image decomposition. Our method relies on the above image transform to suppress surface shading variations, and perform probabilistic reflectance clustering on the flattened image instead of the original input image to achieve higher accuracy. Extensive testing on the Intrinsic-Images-in-the-Wild database indicates our method can perform significantly better than existing techniques both visually and numerically. The obtained intrinsic images have been successfully used in two applications, surface retexturing and 3D object compositing in photographs. Download Comparison Results of L1 smoothing and Intrinsic decomposition Figure 2   Comparison between our piecewise flattening and existing edge-preserving smoothing methods on a 1D signal.         Figure 3   Comparison of intrinsic image decomposition results from our method and other state-of-the-art methods. Acknowledgements We wish to thank Nicolas Bonneel for sharing the binary code of [Bonneel et al. 2014] and the anonymous reviewers for their valuable comments. This work was partially supported by Hong Kong Research Grants Council under General Research Funds (HKU719313). Bibtex @article{BiHY15,         author = "Bi, S. and Han, X. and Yu, Y.",         title = "An \emph{L}$${}_{\mbox{1}}$$ image transform for edge-preserving smoothing and scene-level intrinsic decomposition",         journal = "{ACM} Transactions on Graphics",         volume = "34",         number = "4",         pages = "78",         month = "",         year = "2015" }