Abstract: Objective: At present, most of beam hardening calibration algorithms required spectrum, the detector characteristics and other information. As the beam hardening calibration algorithm's complexity, we propose a method based on image post-processing. Methods: Firstly, the water portion with cupping artifacts is segmented in the water phantom CT image and done Radon transform to obtain projection information. The calibration model with unknown coefficients is built with above projection information. Secondly, an ideal water phantom image without beam hardening artifacts is constructed and compared with calibration model. When the difference between them is minimal, the unknown coefficients of correction model is determined. Finally, use the calibration model that combined the determined parameter and the projection information of segmented soft tissue to correct the practical CT images so that eliminate the cupping artifacts. Results: The result of computer simulation shows that this method can efficiently remove the cupping beam hardening due to a medical CT beam-hardening caused. Conclusion: The proposed method without need for projection data and spectrum information can be carried out in beam hardening cupping artifacts calibration, improved beam hardening correction flexibility.