作者: Tubagus Maulana Kusuma , Hans-Jürgen Zepernick
关键词: Computer vision 、 Computer science 、 JPEG 、 Metric (unit) 、 Human visual system model 、 Image quality 、 Artificial intelligence 、 Hybrid image 、 Quality (business) 、 Service quality 、 Subjective video quality
摘要: User-oriented image quality assessment has become a key factor in multimedia communications as means of monitoring perceptual service quality. However, existing metrics such Peak Signal-to-Noise Ratio (PSNR) are inappropriate for in-service since they require the original to be available at receiver. Although PSNR and others objective metrics, not based on human visual perception typically designed measure fidelity. On other hand, system (HVS) is more sensitive than In order overcome these problems, we propose novel reduced-reference hybrid metric (RR-HIQM) that accounts does reference This combination several artifact measures. The result single number, which represents overall