With the emergence of counter-stains, a multi-spectral imaging method based on the multi-channel narrow-band filter and color image sensor was proposed in this paper. Firstly, a multi-channel narrow-band filter was used to separate the light source. Then, the multi-spectral stripping model was established by using the multiple linear regression method to separate the monochromatic light aliasing response from the color image sensor. The multi-spectral image is real-time delaminated and output by FPGA(Field Programmable Gate Array). The experimental results show that, the Full Width at Half Maximum(FWHM) of the multi-channel narrow-band filter is 12 nm/15 nm/20 nm, and the transmission light is really a narrow-band monochromatic light in three different wavelengths; the model has high accuracy when the test level is 0.01; the substance with the same spectral characteristics has a consistent feature on the same band grayscale image and the same substance has an obvious different feature on different bands grayscale image.