Noise Specific Blind Image Quality Assessment Using Three Statistical Features

This paper presents a blind image quality assessment (BIQA) method for noise distorted images. In the proposed method, three different domains are simultaneously utilized and three statistical features are extracted from the input noisy image. The noise distortion types to be covered are more widened than those done by the conventional BIQA methods. Experiments demonstrate that the proposed method achieves more accurate BIQA from noisy images than the state-of-the-art methods do.