In circumstance when there are no pure pixels exits in hyperspectral image

the endmembers extracted by traditional algorithms are usually mixing ones still. In order to solve the problem

this paper proposes an endmember extraction algorithm based on the re-analysis of near-endmember using the linear mixing model. After extracting the pixels which were most close to the pure endmembers in the image

using the convex polyhedron’s geometric characters the algorithm searches the pixels around the near-endmember and finds out the pixels in the edge of the convex polyhedron formed by the pure endmembers. Calculating the mixing coefficients of every endmember in these pixels by the laws of sins

thus

with these coefficients the pure endmember vectors could be extracted according to the constraints of linear mixing model. A hyperspectral image simulated by the real spectrums is used to investigate the performance of the algorithm. Preliminary result indicates the effectivity of the algorithm. Applying the algorithm to a real Hyperion image it can also get a better result.