计算机技术与发展2017，Vol.27Issue(8)：20-24,29,6.DOI:10.3969/j.issn.1673-629X.2017.08.005
针对SAR图像的树形稀疏表示结构识别算法研究
Investigation on Identification Algorithm of Tree-structure Sparse Representation for SAR Target
摘要
Abstract
In order to improve the ability of identifying SAR target series with sparse representation,a tree-structure sparse coding recognition algorithm is proposed,which is employed to lift the recognition accuracy of target models.The sparse representation tree is a tree-like classifier composed of multiple nodes,each of which has a sparse representation dictionary and a classifier for the node.The sparse representation algorithm is used to solve the eigenvector of unknown sample on a single node,realizing the target type identification according to the minimum principle of reconstruction error.The root node is employed to direct input SAR images with similar sparse results to children nodes,which have more specialized dictionaries and classifiers to identify these target series.Experiments on MSTAR target dataset show that it is suitable for the sample distribution and has improved target recognition rate up to 84%,and that compared with the traditional sparse coding method,it has got effective improvement on the target series recognition accuracy without more time expenditure.关键词
SAR目标识别/型号识别/树形信息字典/稀疏表示/字典学习Key words
SAR automatic target recognition/series recognition/tree-structure information dictionary/sparse representation/dictionary learning分类
信息技术与安全科学引用本文复制引用
陈春林,张礼,刘学军..针对SAR图像的树形稀疏表示结构识别算法研究[J].计算机技术与发展,2017,27(8):20-24,29,6.基金项目
中国航空科学基金(20151452021,20152752033) （20151452021,20152752033）
