Expert pattern recognition assessment of image quality

The authors propose a method for objectively determining video transmission quality that is user-oriented and independent of coding/decoding and transmission technologies. The technique uses an expert pattern recognition system that operates on features of a received image with unknown transmission impairments. An example of a candidate feature is provided. The system provides quality classification on a minimum probability-of-error basis relative to subjective classification. The expert system uses statistical pattern recognition to perform low-level processing and an initial classification. Additional knowledge about subjective classification, visual perception, user requirements, and application bias can be applied by the expert system to refine the classification