There has been a significant increase in video contents in Internet traffic. However, it is very hard to guarantee QoS for streaming in Internet due to the time varying channel characteristics of Internet . To overcome these problems in the application layer of OSI-7 layer, MPEG-4 Fine Granular Scalability(FGS) has been standardized recently. MPEG-4 FGS can simply adapt to channel bandwidth variations in real-time. And FGS selective enhancement(SE) has been adopted in the MPEG-4 FGS standard to enhance the subjective quality.
This paper presents a rectangular region of interest(ROI) based simplified SE technique that enables the FGS scheme to perform ROI-based enhancement and prioritized transmission of ROI region. The proposed method enhances the subjective quality of the rectangular ROI with significantly reduced overhead compared with the original SE. The proposed algorithms can be characterized by two new techniques. The first is an efficient selection of the ROI using the spatio-temporal correlations existing in moving pictures. And the second is to reduce most of the overhead bits required for SE by reformatting and simplifying the original FGS syntax. It is shown by simulation that the proposed SE can provide a good visual quality for the ROI with significantly reduced overhead bits
