文献类型：期刊文献

英文题名：Automatic detection and removal of color comments from book images

作者：Dai, Guang-Hui[1]; Yuan, Yu-Bo[1]; Zhang, Jing[1,2]; Chen, Zhi-Hua[1]

机构：[1] Department of Computer Science and Engineering, East China University of Science and Technology, Shanghai, 200237, China; [2] State Key Lab. for Novel Software Technology, Nanjing University, Nanjing, 210093, China

年份：2015

卷号：2

起止页码：591

外文期刊名：Proceedings - International Conference on Machine Learning and Cybernetics

收录：EI(收录号：20172403790171)

语种：英文

外文关键词：Gaussian distribution

摘要：Locating the salient parts or regions is a very challenging and important step in image processing. It will be very helpful for image denoising, inpainting or smoothing. In this work, we propose a novel algorithm to detect the regions with color comments in book images automatically. A truncated function is employed to locate positions with high values of Gaussian curvature. Each channel information of a book image is treated as a surface on a given square and corresponding values of the Gaussian curvature are calculated. Finally, with the training cutoff, pixels with high Gaussian curvature will be removed from the original image and replaced by white points. Experimental results ensure that the proposed algorithm is very effective and removes comments automatically. ? 2015 IEEE.