In the process of X-ray computed tomography (CT) imaging, the traditional single-energy X-ray CT imaging technology is only applicableto structural analysis but can’t meet the needs of functioning for substance distinction and identification because of the multispectral hardening artifacts and inconsistency between the projection acquisition process and reconstruction assumption. A multispectral CT imaging method based on the spectrum matching priors is presented. First, energy spectrum filtering matching model is built and range spectrum parameters are set according to the material composition; then multi-spectrum projection sequence is acquired by filtering. Second, the different reference energy is selected according to the material composition, befor using the improved Algebraic Reconstruction Techniques (ART) to achieve a multi-spectral CT imaging. Simulation result shows that we improved the contrast of the reconstructed image effectively, while meeting the needs of substance distinction. The actual data collection process is achieved by reconstructing in different spectrum and different corresponding reference energy.