Scoliosis diagnosis relies heavily on CT imaging, but existing filtering methods often struggle to balance noise reduction and edge preservation. To address this issue, this paper proposes a novel hybrid filtering approach based on the Huber penalty function, which combines median filtering and partial histogram equalization. The method enhances image quality by effectively reducing noise while preserving structural details and edge clarity. Evaluations demonstrate the proposed method’s superiority over conventional techniques. Moreover, the enhanced image quality improves the accuracy of Cobb angle measurements, reducing manual annotation errors. These findings highlight the potential of this approach for clinical diagnosis of scoliosis.