As one of the most important payloads on Chang’E-1 satellite

interference imaging spectrometer（IIM） has acquired large amount of hyperspectral data.However

it is still a great challenge to retrieve minerals distribution information of the lunar surface due to the limited spectral wavelength range of IIM data（480—960 nm） which does not cover spectral absorption features of lunar minerals completely.At present

precise quantitative mineral distributions or credible surveys of lunar surface are lacking to validate the retrieved results.In this paper

we perfomed simulation of IIM spectral data based on spectral linear mixing method with four main lunar minerals（plagioclase

clinopyroxene

olivine and ilmenite）.We analyzed the accuracy of mineral endmembers extraction and implementeda preliminary application with true IIM data.We simulated the spectra of lunar mare and highland according to different proportion of plagioclase and ilmenite respectively.We added random noises of the same trend as IIM and employed four endmember extraction approaches(vertex component analysis

independent component analysis

minimum volume simplex analysis

simplex identification via split augmented lagrangian to extract endmembers.We used spectral angle distance（SAD） as a criteria for accuracy analysis.The results show that all the SADs of endmembers extracted by MVSA and SISAL are lower than 0.1.Plagioclase（the SADs that are lower than 0.015） can always be extracted by all approaches.All SADs of clinopyroxene and ilmenite are lower than 0.1 with olivine having the lowest extraction accuracy.