Rectangular-Normalized Superpixel Entropy Index for Image Quality Assessment.

摘要: Image quality assessment (IQA) is a fundamental problem in image processing that aims to measure the objective of distorted image. Traditional full-reference (FR) IQA methods use fixed-size sliding windows obtain structure information but ignore variable spatial configuration information. In order better multi-scale objects, we propose novel method, named RSEI, based on perspective receptive field and entropy. First, find consistence relationship exists between fidelity human visual individuals. Thus, reproduce system (HVS) semantically divide into multiple patches via rectangular-normalized superpixel segmentation. Then weights each are adaptively calculated their volume. We verify effectiveness RSEI by applying it data from TID2008 database denoise algorithms. Experiments show outperforms some state-of-the-art algorithms, including (VIF) weighted average deep (WaDIQaM).