自适应结构化背景和形状特征子空间高光谱图像多类目标检测  

MULTICATEGORY TARGETS DETECTION OF HYPERSPECTRAL IMAGERY BASED ON ADAPTIVE STRUCTURED BACKGROUND AND SHAPE-FEATURE SUBSPACE

作　　者：贺霖[1] 潘泉[2] 邸韡[2] 

机构地区：[1]华南理工大学自动化科学与工程学院,广东广州510641 [2]西北工业大学自动化学院,陕西西安710072

出　　处：《红外与毫米波学报》2007年第5期353-358,共6页Journal of Infrared and Millimeter Waves

基　　金：国家自然科学基金重点项目(60634030);国家自然科学基金(60475004;60602056;60372085);航空科学基金(2006ZC53037);教育部新世纪人才基金(NCET-04-0816);广东省自然科学基金团队项目(04205783);遥感科学国家重点实验室开放基金(SK050013)

摘　　要：针对高光谱图像中目标形状特征已知,背景和目标光谱特征未知时的多类小目标检测问题,给出一种检测算法.通过高光谱图像数据样本二次型的高阶矩控制点扩散函数,获取自适应结构化背景;然后,利用目标形状先验信息构造形状特征子空间,在高维光谱特征空间实现形状特征子空间匹配检测.理论分析和实验结果表明该检测器可同时有效检测具有不同形状特征的多类目标.A new detection algorithm was presented to detect muhicategory targets of known shape-feature and unknown spectral signature in unknown environment. Firstly, a point spread function was constructed via high-order moments of quadratic form of data samples to obtain adaptive structured background. Then, a priori shape-features of targets were utilized to construct a shape-feature subspace which is matched with high-dimension spectral signature space. Theoretic analysis and the results of experiment verify the effectiveness of the algorithm.

关 键 词：信息处理技术 高光谱图像 多类目标检测 形状特征子空间 结构化背景 

分 类 号：P391.4[天文地球—地球物理学]