Perceptually relevant error classification in the context of picture coding

Distortions resulting from typical source coding and channel interference are analyzed for monochrome still pictures and evaluated based on a segmentation-based error metric (SEM). The error picture is segmented into errors at own edges, errors representing exotic or spurious edges, and errors in flat regions to describe edge errors like blurring, exotic structures like blocking and contouring, and residual errors like random noise, respectively. Error parameters or distortion factors are derived by appropriate summation over the segmented components and combined to build the SEM using the impairment addition law by the British Post Office. For a picture data base consisting of typical coding distortions, this leads to a more satisfactory result than a generalized linear summation. Observers are found to be more sensitive to small visible exotic/spurious edges, but less sensitive to edge errors. Yet, large edge errors may rapidly deteriorate the picture quality.