Digital video systems exhibit fundamentally different artifacts than analog video systems. Lossy compression methods such as MPEG introduce distortions that highly depend on scene content. Traditional signal quality measurements are inadequate for the evaluation of the visibility of these artifacts. Subjective methods, undoubtedly the "ultimate truth" about perceived quality, are complex and time-consuming. Simple distortion measures like mean-squared error, on the other hand, albeit popular, do not correlate well with perceived quality.
These problems necessitate advanced automatic methods for video quality assessment. Ideally, such a quality assessment system would perceive and measure video impairments just like a human being. Therefore we focus our research on metrics based on models of the human visual system, which are the most general and potentially the most accurate ones.
The vision model we have been working on incorporates many aspects of human vision, including color perception and a multi-channel architecture. It can accurately model perceptual phenomena such as contrast sensitivity, masking and facilitation. Our video quality assessment system has also been submitted to the Video Quality Experts Group (VQEG) as a proposal in an effort to come up with an international ITU recommendation for a standard in visual quality assessment of video.
Applications of video quality assessment include:
- Evaluation and comparison of video codecs;
- End-to-end testing of transmission systems;
- Online quality monitoring and control;
- Assessment of specific video features;
- Perceptual video restoration.