Hyperspectral remote sensing image provides not only abundant spectral information but also spatial detailed infor-mation.For making full use of the information

this paper considered the variations of the spectral feature among pixels in local neighborhood and researched on a kind of Local Neighborhood Spectral Similarity Measure Index to extract the spatial detailed information.Furthermore

the spectral feature of endmember in local neighborhood was also considered.And a kind of Local Neighborhood Independent Endmember Index was proposed.In experiments

these local neighborhood indexes demonstrated excellent performance in the real hyperspectral image.Based on these work

it can further improve the capability of hyperspectral target detection and identification by combining with spatial and spectral information.