Efficient predictor of subjective video quality rating measures

摘要

An efficient predictor for subjective video quality rating measures processes a reference video and an impaired video corresponding to the reference video in respective processing channels. Each channel applies a contrast sensitivity function from a human vision system model to the respective videos and a contrast discrimination function for self-masking to the output of the contrast sensitivity function. The outputs from the channels are input to a differencing circuit to obtain absolute difference values. The absolute difference values are modified by a global masking factor derived from the output from the reference video channel, and the resulting values are converted into units conforming to a standard quality rating metric.