期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2000
卷号：XXXIII Part B7(/1-4)
页码：776-781
出版社：Copernicus Publications
摘要：Unmixing of mixture pixel is a means of determining the relative abundances o f materials depicted in hyperspectral imagery based on the materials' spectral characteristics. In linear spectral unmixing theory, the reflectance at each pixel of the image is assumed to be a linear combination of the reflectance of each material (or endmember) present within the pixel. However, in many situations, the endmember spectra are imported fro m a spectral library or field spectrometer. In order to complete unmixing calculations, these endmember spectra must be pre-processed and convolved to the bandwidths of imaging spectromter in advance. It is very inconvenient and low efficient for the image unmixing calculation. In this paper, authors have developed a new unmixing algorithm based on the linear spectral unmixing theory. The algorithm employs the chromatic characteristics of endmember spectra to unmix the mixture pixels in the hyperspectral remote sensing image. Advantages of the algorithm are the number of required parameters in unmixing calculations is greatly decreased, the speed of unmixing calculation is much faster. The results of unmixing calculation of the algorithm are satisfactory co mpared with original unmixing method. It is demonstrated the unmixing algorithm developed in this paper can be effectively applied in the classificatio n of imaging spectrometer data
