作者: Hantao Liu , Ingrid Heynderickx
DOI: 10.1109/ICASSP.2008.4517747
关键词:
摘要: A novel no-reference blockiness metric that can automatically and perceptually quantify blocking artifacts of DCT coding is presented. The proposed built upon the specific structure information artifact itself combined with properties human visual system (HVS) by means a simple efficient model masking. Investigations are conducted to reduce additional cost introduced vision model, without compromising its overall prediction ability. validated through comparing performance state-of-the-art HVS based metrics respect accuracy, reliability computational complexity.