抄録
When the agricultural area is evaluated by the remote sensing image, crop and soil can be considered taking various coverages within a pixel by the growth stage of crop. If the large deviation to the coverage of crop and soil is seen, it was possible to estimate endmembers by the conventional singular value decomposing method. However, if the large deviation to the coverage of crop and soil is not seen, it is difficult to estimate endmembers by the conventional singular value decomposition method. In this paper, by using not only the singular value decomposition method but also three restricted conditions and maximum-minimum average, it became possible to estimate endmembers and coverages more precisely also in such cases.
