Abstract: In conventional computed tomography (CT) systems, X-ray beams used for imaging are polyenergetic. These beams cause hardening artifacts in the reconstructed images owing to their polychromatic nature. For a multi-material imaging object, prior information about the X-ray spectrum and composition of the imaging object is required for correcting the beam hardening. To avoid relying on such prior information, this study proposes a beam-hardening artifact correction method for CT imaging of multi-material objects under single voltage. In this method, the X-ray image is considered as a non-negative weighted sum of multiple single-energy X-ray images. The decomposition model for the X-ray transmission images is constructed using the maximum likelihood function of a Gaussian distribution as the objective function under the constraint of the projection integral invariance at different angles. To solve this model, the projection of the spatial terms on the X-ray path is obtained. Thus, CT images for X-ray beams with different energies can be reconstructed. For imaging experiments with dual-material and multi-material models, compared with directly reconstructed CT images, the hardening artifacts in the reconstructed images were significantly reduced using this method, indicating its effectiveness.