基于集合论的光谱相似性度量及在影像检索中的应用

Spectral Similarity Measure Based on Set Theory and Its Application to Hyperspectral RS Image Retrieval

将高光谱遥感中光谱向量相似性度量转换为相应的集合相似性度量,提出了两种适用于光谱的相似性度量方法,即基于光谱多边形面积和基于特征波段位置匹配度量.实验结果表明,基于光谱多边形的方法能够更有效地综合应用反射率和波长二维信息,有效度量光谱向量相似性.

Based on analysis to the principles and indexes of similarity measure by set theory and set operation, the similarity measure to spectral vectors in hyperspectral RS image was transformed into the similarity measure to corresponding sets. Two approaches used to spectral similarity measure based on set theory were proposed, one is similarity measure by the area of spectral polygon, and the other is by locations of characterized bands. It proved that the approach based on spectral polygon can be used to similarity measure and image retrieval effectively because of its synthetic applications to albedo and wavelength information.