Volume 16 - Issue 1
Rectangular Normalized Superpixel Entropy Index for Image Quality Assessment
Abstract
Full-reference picture quality appraisal calculations ordinarily perform examinations of highlights separated from square fixes. These patches don't have any visual implications. In actuality, a super pixel is a lot of picture pixels that offer comparable visual qualities and is along these lines perceptually important. Highlights from super pixels may improve the presentation of picture quality appraisal. Enlivened by this, we propose another super pixel-based comparability file (SPSIM) by separating perceptually important highlights and amending likeness measures. The proposed technique assesses picture quality based on three estimations, in particular, super pixel luminance similitude, super pixel chrominance likeness, and pixel angle comparability. The initial two estimations evaluate the general visual impact on neighborhood pictures. The third estimation evaluates auxiliary varieties. The effect of super pixel-put together territorial slope consistency with respect to picture quality is additionally broke down. Misshaped pictures indicating high local inclination consistency with the relating reference pictures are outwardly valued. In this manner, the three estimations are additionally updated by fusing local angle consistency into their calculations. A weighting capacity that demonstrate super pixel based surface unpredictability is used in the pooling stage to get the last quality score. Tests on a few benchmark databases show that the proposed strategy is serious with cutting edge measurements.
Paper Details
PaperID: 201019
Author's Name: V. Lalithambica, K. Praveen Kumar and P. Siva Prasad
Volume: Volume 16
Issues: Issue 1
Keywords: Full-reference, Picture Quality Appraisal, Super Pixel, Provincial Slope Consistency, Surface Multifaceted Nature.
Year: 2020
Month: February
DOI: DOI_name
Pages: 121-