Output details

A novel perceptual quality metric for video compression

This paper was presented at the Picture Coding Symposium, which is the pioneer international forum devoted to visual data coding. This perceptual quality metric is based on the Mean Squared Error (MSE), which can be efficiently calculated, and the visual activity dependant correlation between MSE and perceptual quality. The visual activity is measured using simple spatial and temporal filters. The simple error and activity calculations make this algorithm particularly suitable for real-time algorithms. This metric consistently out-performs well established metrics like VSSIM. This work lead to further development of the metric based on this principle, which resulted in a new algorithm that out-performs well established popular perceptual quality metrics.