HVS-based image quality assessment for digital cinema

Improving image quality assessment with modeling visual attention

“…The third approach, based on local block quality, is better than the first and the second approaches in most cases, especially for Digital Cinema setup. We believe the reason is that Digital Cinema always has a large size screen, so that subjects can not see an entire image at once [11]. Compared to the use of an entire attention map, using a subset of the attention map usually has better performance, except for the Saliency model using the second approach.…”

“…Three respective datasets were employed, including LIVE image quality database [14], stereoscopic image quality database [15], and an image quality dataset in DC setup [11].…”

“…We have found that contrast information is an important factor in image quality assessment [11]. Human usually pay more attention to those regions that have higher contrast levels.…”

“…However, Liu et al [10] reported that visual attention can be beneficial for two objective image quality metrics based on "ground truth" visual attention data from an eye-tracking experiment on natural images. We also found that visual attention detection can improve the accuracy in predicting video quality with general degradation [8] and image quality assessment in Digital Cinema (DC) scenario [11].…”

Exploring alternative content in Digital Cinema

“…We participate in the research of QoE in DCinema by exploiting a commercial DCI-specified cinema in Trondheim (Nova Kino 1) as a realistic test environment for numerous experiments, and we had made several contributions on visual perceptual quality of high quality imagery in DCinema [7], [8], [9].…”