Video quality assessment using objective parameters based on image segmentation

This work presents a new methodology for video quality assessment using objective parameters based on image segmentation. The method is applied to a set of natural scenes and MPEG-2 video codecs, An image segmentation algorithm divides the scenes into three regions: plane, edges, and texture; and assigns a set of objective parameters to each region, A perceptual-based model is defined by computing the relationship between objective measures and results of subjective assessment tests. Thereafter the model allows subjective rating prediction without human input. The work shows that region-based objective measurements provide more accurate predictions as compared to global measurements.