Aiming at the problem of explosive powder detection on suspected personnel and clothing surfaces in a wide open space, a remote sensing method of explosive powder surface contamination based on NIR spectroscopy was studied, a NIR imaging spectral data acquisition system was developed, the NIR reflection characteristic spectra of various explosive powder and contamination substrates was measured, numbers of explosive powder surface contamination samples were prepared. In view of the aliasing problem for NIR reflection characteristics of explosive powder and substrate, a NIR spectral unmixing correction model was constructed by using NIR spectral data processing technology to remove the interference of contaminated substrate signal on the identification of explosive powder. Aiming at the interference caused by uneven illumination of the light source(saturation due to strong light reflection and weak signal by shadow), the correction score maps were effectively filtered to avoid misidentification problems. In addition, the problem of false identification caused by excessive spectral pretreatment with large background noise was corrected by using the mean spectral reflectance and score maps. The experiments show that the problem of surface contamination aliasing is solved, the interference of illumination and other noise factors are removed, the misclassification is avoided, AP (ammonium perchlorate), CL-20 (hexanitrohexanithine), NQ (nitroguanidine), RDX (blacksorkin), TATB (triaminotrinitrobenzene), Nidi (industrial explosives), fireworks and other explosive powders and mixtures are successfully identified on the substrates of typical background materials (Cotton and linen cloth, chemical fiber cloth), the feasibility of the system and method is verified, the remote sensing imaging alarm of explosive powder surface contamination is realized first in the laboratory, and the effective distance can reach tens of meters, the system has certain application value and development potential.