A method of image quality assessment based on region of interest

We apply a method of image quality assessment based on structure similarity (SSIM) to images including regions of interest (ROIs). A gradient-based SSIM model that takes the gradient as the main structural information is used among ROIs, while the classical SSIM model is followed in other regions. Then the final image quality index is evaluated after weights-taxed averaging of indexes in every region. Experimental results show that, considering the importance of ROIs sufficiently, our model is more consistent with perceived visual quality.