CE8ROI-basedScalableVideoCoding.docVIP

CE8: ROI-based Scalable Video Coding Coordinator: Institute for Infocomm Research (Zhongkang Lu) Proponent (Company, Contact Person, Email) Verification Partner (Company, Contact Person, Email) National Chiao-Tung University Wen-Hsiao Peng pawn@ ICU, ETRI Truong Cong Thang tcthang@icu.ac.kr Haechul Choi choihc@etri.re.kr ICU, ETRI Truong Cong Thang tcthang@icu.ac.kr Haechul Choi choihc@etri.re.kr Institute for Infocomm Research Zhongkang Lu zklu@.sg Institute for Infocomm Research Zhongkang Lu zklu@.sg Panasonic Singapore Laboratories Shen Shengmei shen@.sg Functionality/Problem Addressed The CE addresses the region-of-interest functionality, which can also achieve progressive refinement and flexible slice structure. In SVC WD1.0, a coarse ROI functionality can be achieved by flexible macroblock order (FMO) and possibly QP adaptation scheme. To fully understand the advantages and disadvantages of the existing approach and possibly improvement offered by using a new approach, we like to do a further investigation on the above issues. The CE will integrate a ROI description into the current JSVM reference software, and investigate different ways to support the ROI-based scalable video coding, including ROI-based fine granular scalability (FGS), ROI-based spatial scalability, and ROI-based coarse granular scalability (CGS). Description of tools Tool1: Selective enhancement functionality to scalable video coding (JVT-O020) Regions-of-interest (ROI) is a desirable feature in some applications as described in the requirement document of MPEG-4 Part 10 Amd. 1 Scalable Video Coding (SVC), N6880. For those applications, the clients at the decoder side may require to receive a better decoded quality in the ROIs when the pre-encoded scalable bit-stream is truncated. Currently, the JSVM1.0 of SVC does not support ROI functionality. In this proposal, we develop a graceful and arbitrary shape selective enhancement scheme to provide the ROI functionality in SVC. To pr