Reduced-Reference Image Quality Assessment Based on Free-Energy Principle with Multi-Channel Decomposition

Abstract

The free-energy principle studied in brain theory and neuroscience accounts for the mechanism of perception and understanding in human brain, which is highly adapted for measuring the visual quality of perceptions. On the other hand, psychologists and neurologists report that different frequency and orientation components of one stimulus arouse different neurons in striate cortex. In this paper, a novel reduce-reference (RR) image quality assessment (IQA) metric based on free-energy principle in multi-channel is proposed, which is called MCFEM (Multi-Channel Free-Energy principle Metric). We first decompose the input reference image and distorted image via a two-level discrete Haar wavelet transform (DHWT). Next, the free-energy features of each subband images are computed based on sparse representation. Finally, an overall quality index is received through the support vector regressor (SVR). Extensive experimental comparisons on four (LIVE, CSIQ, TID2008 and TID2013) benchmark image databases demonstrate that the proposed method is highly competitive with the representative RR and no-reference models as well as full-reference ones.