%0 Journal Article
%A HU Gen-sheng
%A LIANG Dong
%A ZHANG Xue-min
%T Infected Pine Recognition in Remote Sensing Images Using WWSVDD Multi-Classification
%D 2014
%R 10.13190/j.jbupt.2014.02.006
%J Journal of Beijing University of Posts and Telecommunications
%P 23-27
%V 37
%N 2
%X <p>For the problem of misjudgment in the fuzzy data domain, a modified multi-classification algorithm of weighted wavelet support vector data description (WWSVDD) is proposed and applied to the infected pine recognition in remote sensing images. Firstly, utilizing the high resolution images acquired by the double spectrum camera fixed on the unmanned aerial vehicle, the features of ground objects are extracted to construct the corresponding feature vectors. Secondly, each kind of ground objects samples is described by WWSVDD model. Finally, according to the distribution of test samples in the feature space, the methods of minimum relative distance and membership function are respectively used to decide the labels of the samples, so that the infected pines are recognized ultimately. The experiment results show that the proposed method can recognize the infected pines more effectively than the traditional multi-classification methods of <em>K</em>-nearest neighbor (KNN) and support vector data description (SVDD).</p>
%U https://journal.bupt.edu.cn/EN/10.13190/j.jbupt.2014.02.006