A novel spatial pooling technique for image quality assessment based on luminance-contrast dependence

The paper presents a spatial pooling technique for image quality assessment (IQA) that is based on the idea that the adaptive mechanisms of luminance and contrast in the early vision operate independently. The work is motivated by recent vision science studies on this topic that have not been investigated yet in the field of IQA. The Structural SIMilarity index (SSIM) has been selected as the base IQA measure due to its explicit utilization of the local luminance mean and contrast of both original and degraded images. Experimental results show that a spatial pooling algorithm that only depends on the degraded image results in significantly improved image quality prediction.