Unsupervised fully constrained least squares linear spectral mixture analysis method for multispectral imagery

Subpixel detection and quantification of materials in multispectral imagery presents a challenging problem due to a relatively low number of spectral bands available for analysis in which case the number of spectral bands may be less than the number of materials to be detected and quantified. The problem is even more difficult when the image scene is unknown and no prior knowledge is available. Under this circumstance, the desired information must be obtained directly from the image data. In this paper, we present an unsupervised least squares-based linear mixture analysis method coupled with a band expansion technique for multispectral image analysis. This method allows us to extract necessary endmember information from an unknown image scene so that the endmembers present in the image can be detected and quantified. The band expansion technique creates additional bands from the existing multispectral bands using band-to-band nonlinear correlation. These expanded bands ease the problem of insufficient bands in multispectral imagery and can improve and enhance the performance of the proposed method. The experimental results demonstrate the advantages of the proposed approach.