The natural characteristics of image signals and the statistics of measurement noise are decisive for designing optimal filter sets and optimal estimation methods in signal processing. Astonishingly, this principle has so far only partially found its way into the field of image sequence processing. We show how a Wiener-type MMSE optimization criterion for the resulting image signal, based on a simple covariance model of images or image sequences, provides direct and intelligible solutions for various, apparently different, problems, such as error concealment, or adaption of filters to signal and noise statistics.
A statistical unification of image interpolation, error concealment, and source-adapted filter design
01.01.2004
367858 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Modified Edge-oriented Spatial Interpolation for Consecutive Blocks Error Concealment
British Library Conference Proceedings | 2005
|Novel Error Concealment Algorithm for Image Transmission in JPEG2000
British Library Online Contents | 2004
|Error Concealment for Motion JPEG2000
British Library Conference Proceedings | 2005
|Error concealment for motion JPEG2000
IEEE | 2005
|