Chan-Vese (CV) model is robust to noise to some extent due to the using of global image information. But for the image corrupted by strong noise, CV model cannot get satisfactory segmentation. In this paper, we propose a variational level set model that combines CV model with variational image decomposition. The proposed model integrating BV-L² decomposition into CV functional can achieve image denoising and segmentation, simultaneously. An alternative and iterative algorithm is applied to numerically solve the proposed model. Experiments on some synthetic and real images demonstrate the efficiency and robustness of the proposed model. Moreover, compared with the well-known CV model and VFCMS model, the proposed model can get better performance for the image corrupted by strong noise.