Synthetically utilizing image invariant moments feature and SVM(Support Vector Machine)classification method,a novel recognition algorithm was proposed to deal with multiview target in infrared images.Firstly,the target silhouette image was extracted by image segmentation method.Then,some proper order moments which are selected from Hu moments,Zernike moments and Fourier-Mellin moments,could be used as the invariant eigen vector for target in the specific view.The angular field of view to targets was divided into several parts for solving the samples variety caused by looking from multi-view angle.In each divided field,pair-wise SVM classifier was used to realize the multi-target classification.A large number of recognition testing on multi-view targets in infrared images proves the validity and reliability of the scheme in this paper.
参考文献
相似文献
引证文献
引用本文
马超杰,李晓霞,杨华,吴丹,王静雯.应用支持向量机分类的多角度目标识别技术[J].激光与红外,2009,39(1):88~91 MA Chao-jie, LI Xiao-xia, YANG Hua, WU Dan, WANG Jing-wen. Multi-view Target Recognition Algorithm Based on Support Vector Machine Classification[J]. LASER & INFRARED,2009,39(1):88~91