This study aims to develop a multi-spectral imaging system based on a fully digital method. This method can reconstruct the spectral reflectance from digital counts of images. Firstly, we setup spectroradiometer and CCD to measure the spectral reflectance and the digital counts of the training dataset. Then, we used Principal Component Analysis (PCA) to reduce both the dimension of database and computational effort. Form the spectral reflectance and digital counts, the system can establish a linear transformation as a mathematical model. Finally, the spectral reflectance can be reconstructed through mathematical model. When we get the digital counts of testing object by a snapshot, we are able to obtain the spectral reflectance. The experimental result shows the color difference is smaller than 3, which means the mathematical model is robust and savvy. Finally, we applied this technology to distinguish the phenomenon of Metamerism. We expect this technology can be applied inspection of bio-medical engineering in the future.