Improving image quality assessment with modeling visual attention

Abstract

Visual attention is an important attribute of the human visual system (HVS), while it has not been explored in image quality assessment adequately. This paper investigates the capabilities of visual attention models for image quality assessment in different scenarios: two-dimensional images, stereoscopic images, and Digital Cinema setup. Three bottom-up attention models are employed to detect attention regions and find fixation points from an image and compute respective attention maps. Different approaches for integrating the visual attention models into several image quality metrics are evaluated with respect to three different image quality data sets. Experimental results demonstrate that visual attention is a positive factor that can not be ignored in improving the performance of image quality metrics in perceptual quality assessment.