ROI-preserving multiple description coding based on pre- and post-processing to standard encoders
Gabriella Olmo, Tammam Tillo · 2003
Multiple description coding of images can be implemented as pre- and post-processing by means of simple splitting of the pixels into subimages. A controllable amount of extra redundancy is inserted in order to improve estimation quality in case of one descriptor reception. This redundancy is usually inserted in a uniform spatial fashion. In this papers we propose a novel pre- and post-processing method which exploits the directional spatial characteristics of a region of interest in order to protect it in a multiple description coding scenario. Simulation results show a noticeable performance with respect to other proposed methods.