Abstract: A novel fidelity constrained P-M model for image denoising is proposed in order to improve P-M model for image denoising. The merits of P-M model and fidelity constrained term are combined, and a fidelity constrained is exploited to adjust smoothing weight, which can remove noise effectively, preserve image characteristics and avoid the staircasing effect. Experimental results demonstrate that the proposed method is effective for image denoising.