文献类型：期刊文献

英文题名：A variable precision rough set approach to the remote sensing land use/cover classification

作者：Pan, Xin[1,2,3] Zhang, Shuqing[1] Zhang, Huaiqing[4] Na, Xiaodong[1,2] Li, Xiaofeng[1]

第一作者：Pan, Xin

通信作者：Zhang, SQ[1]

机构：[1]Chinese Acad Sci, NE Inst Geog & Agr Ecol, Changchun 130012, Peoples R China;[2]Chinese Acad Sci, Grad Univ, Beijing 100039, Peoples R China;[3]Changchun Inst Technol, Sch Elect & Informat Technol, Changchun 130012, Peoples R China;[4]Chinese Acad Forestry, Inst Forest Resources Informat, Beijing 100091, Peoples R China

年份：2010

卷号：36

期号：12

起止页码：1466-1473

外文期刊名：COMPUTERS & GEOSCIENCES

收录：;EI(收录号：20104913468645);Scopus(收录号：2-s2.0-78649797191);WOS:【SSCI(收录号:WOS:000286304000001)，SCI-EXPANDED(收录号:WOS:000286304000001)】；

基金：This research was supported in part by the National Natural Science Foundation of China (No. 40871188), the Knowledge Innovation Program of Chinese Academy of Sciences (ZCX2-YW-Q10-1-3). We thank Prof. Patricia Dale of Griffith School of Environment of Australia Griffith University, for her careful revision and fruitful comments on the manuscript.

语种：英文

外文关键词：Remote sensing classification; Knowledge discovery; Overlapping data; Variable precision rough sets; VPRS

摘要：Nowadays the rough set method is receiving increasing attention in remote sensing classification although one of the major drawbacks of the method is that it is too sensitive to the spectral confusion between-class and spectral variation within-class. In this paper, a novel remote sensing classification approach based on variable precision rough sets (VPRS) is proposed by relaxing subset operators through the inclusion error beta. The remote sensing classification algorithm based on VPRS includes three steps: (1) spectral and textural information (or other input data) discretization, (2) feature selection, and (3) classification rule extraction. The new method proposed here is tested with Landsat-5 TM data. The experiment shows that admitting various inclusion errors beta, can improve classification performance including feature selection and generalization ability. The inclusion of beta also prevents the overfitting to the training data. With the inclusion of beta, higher classification accuracy is obtained. When beta=0 (i.e., the original rough set based classifier), overfitting to the training data occurs, with the overall accuracy=0.6778 and unrecognizable percentage=12%. When beta=0.07, the highest classification performance is reached with overall accuracy and unrecognizable percentage up to 0.8873% and 2.6%, respectively. (C) 2010 Elsevier Ltd. All rights reserved.