Abstract: To address the limitations of the adaptive L^p norm-based image restoration (ALPIR) model, an image restoration model based on adaptive local constraints is formulated. Though the ALPIR model can restore images successfully, the staircase effect appears in the flat regions of the restored images, which is closely associated with the poor scheme of the L^p norm of the ALPIR model. Given the deficiency, a new scheme is proposed according to the perceptive characteristic of the human visual system. This scheme has the same form as the visibility function, and applies the orientation information measure and local variance to the detection of edges. The proposed model is applied to image denoising and deblurring, and experimental results demonstrate that it outperforms the ALPIR model, i. e., the edge details of an image are preserved when the image is restored and the staircase effect is depressed by using the model.