Learning a Blind Measure of Perceptual Image Quality

To appear at CVPR 2011

Huixuan Tang#      Neel Joshi$      Ashish Kapoor$     


#University of Toronto


$Microsoft Research


Abstract

It is often desirable to evaluate an image based on its quality. For many computer vision applications, a perceptually meaningful measure is the most relevant for evaluation; however, most commonly used measure do not map well to human judgements of image quality. A further complication of many existing image measure is that they require a reference image, which is often not available in practice. In this paper, we present a “blind” image quality measure, where potentially neither the groundtruth image nor the degradation process are known. Our method uses a set of novel low-level image features in a machine learning framework to learn a mapping from these features to subjective image quality scores. The image quality features stem from natural image measure and texture statistics. Experiments on a standard image quality benchmark dataset shows that our method outperforms the current state of art.

Results

   

Paper (2.29 MB)




Copyright Microsoft Corporation 2010