文献类型：期刊文献

英文题名：A novel matrix-pattern-oriented Ho-Kashyap classifier with locally spatial smoothness

作者：Wang, Zhe[1]; Chen, Songcan[2]; Gao, Daqi[1]

机构：[1] Department of Computer Science & Engineering, East China University of Science & Technology, Shanghai, 200237, China; [2] Department of Computer Science & Engineering, Nanjing University of Aeronautics & Astronautics, Nanjing, 210016, China

年份：2009

卷号：56

起止页码：441

外文期刊名：Advances in Intelligent and Soft Computing

收录：EI(收录号：20150800544172)

语种：英文

外文关键词：Classification (of information)

摘要：The previous work matrix-pattern-oriented Ho-Kashyap classifier (MatMHKS) can directly deal with images in matrix representation n1 × n2 such that the spatial information within these images is not destroyed. Although MatMHKS works with n1 × n2 per image, it is far less that this spatial correlation with the matrix form n1 × n2 can suggest the real number of freedom. MatMHKS just keeps the relationship of the pixels in the same row or column of images. In this paper we further consider the relationship of the pixels that close to each other may be correlated, and thus develop a new matrix-pattern-oriented Ho-Kashyap classifier named MatHKLSS that is introduced with a locally spatial smoothness. The experimental results here demonstrate that the proposed MatHKLSS has a superior advantage to MatMHKS in terms of classification. ? Springer-Verlag Berlin Heidelberg 2009.