As adoption of HEVC has slowed, companies are taking another look at what additional efficiencies can be gained with their existing H.264-compliant encoders. Evaluating the benefits of different encoding algorithms, to determine how many bits can be saved while maintaining the same video quality, requires an effective way of measuring video quality.
Subjective testing that uses human subjects to view and score videos in a controlled environment has always been the gold standard for video quality measurement. However, because subjective testing is both expensive and slow, analysts have turned to objective metrics to provide faster and cheaper measures of video quality.
Historically, PSNR (Peak Signal to Noise Ratio) has been the most widely-used objective metric, but PSNR has well-known limitations. PSNR reflects the mean-squared error over an entire frame, but different parts of a frame can be perceived differently by the human visual system (HVS). More recently, a number of other objective metrics have been proposed to approximate the human perception, but no single objective metric has yet been adopted by the industry.
EuclidIQ has been developing an innovative compression technology called IQ264 that applies perceptual quality optimization (PQO) that integrates consideration of the HVS to improve standard H.264 encoding. Because IQ264 optimizes human perception, its gains are best measured through subjective testing. For this purpose, a subjective testing approach was needed that would produce meaningful results without being prohibitive in terms of time or cost.
The white paper presented here details the resulting subjective test methodology that provides a way to:
– Prove the benefits of the IQ264 PQO technology to current and potential customers; and
– Measure the overall compression benefit of IQ264 as the PQO technology is developed and productized.
The ideas in the white paper can be modified and tailored to specific application needs, resulting in alternative subjective test methodologies.
The white paper, Subjective Test Methodology Design for Perceptual Quality Optimization, can be viewed for free here.