In order to solve the problem in the application of the current multi-spectral imaging device, the neural network algorithm was applied in multi-channel image color reconstruction. With the high-efficiency conversion model of color information of multi-channel images in the multi-spectral imaging system consisting of CCD camera and LED light source, a neural network conversion model under D50 light source was established and color reconstruction experiments using BP neural network and GRNN neural network were carried out. The result showed that the built model could obtain higher chroma reconstruction accuracy and the color reconstruction performance of the replacement target sample remained stable.