Cervical cancer ranks as the fourth most widely detected cancer globally and stands as the fourth highest contributor to cancer-related deaths in women worldwide. If detected early and treated effectively, cervical cancer can be cured. Using automated computer-aided diagnostic (CAD) systems for diagnosing cervical cancer requires high-quality images of cells from the cervix. Due to factors such as differences in image acquisition equipment, the colors of the acquired cervical cell images vary greatly. Moreover, the quality of cervical cell images significantly influences the diagnosis of cervical cancer. Therefore, a network based on improved cycle-consistent generative adversarial network (CycleGAN) is proposed to perform the cervical cell image color correction task. The proposed method modifies the generator of CycleGAN and introduces structural similarity (SSIM) loss in the loss function, which enhances the effectiveness of the proposed method. The results of the experiments indicate that the method we propose not only improves color correction but also enhances the details of cervical cell images.