A study on an ROI/xROI based rate control scheme in advanced video coding for video telephony applications

Rate control is one of the most important schemes in video coding for constant bit rate (CBR) applications such as broadcasting and video telephony. 3GPP and 3GPP2 have adopted for video telephony services a recent video coding standard, H.264|MPEG-4 Part 10 Advanced Video Coding (AVC). The video telephony applications usually require video encoders to be designed to yield their output bitstreams at target bitrates. Since the channel bandwidth in 3G communication networks is very much limited for video telephony services, the overall quality of video data can be degraded in case of directly using an existing rate control algorithm recommended for AVC. Therefore, it is worthwhile to enhance a subjective quality of video contents via ROI based coding while controlling the output bit-rate. A facial region in a scene can be more importantly treated as regions of interest (ROI). However, a simple ROI based rate control which merely divides a frame into ROI and non-ROI is not enough to satisfy users in regard with the quality and does not consider the variation of output bit-rate. In this thesis, an ROI and xROI (extended ROI) based rate-control method is studied, which considers the coding of both the ROI and the xROI``s in non-ROI to meet given target bitrates. In the proposed method, the QP values are increasingly assigned in MB wide inside the non-ROI away from ROI. This reduces the abrupt change in visual quality and the amounts of residual signals along the border between ROI and non-ROI. In this regard, the subjective visual quality is enhanced as well as the proposed rate control has flexibility to control the amounts of the output bit-streams. Besides, two approaches for controlling the amount of encoded bits are introduced to reduce the fluctuation of output bit-rate per GOP. One approach is a deterministic scheme using a look-up table which contains QP changes given some conditions. The other approach is an adaptive scheme based on gradient descent met...
Advisors
Kim, Mun-Churlresearcher김문철researcher
Publisher
한국정보통신대학교
Issue Date
2009
Identifier
393081/225023 / 020074246
Language
eng
Description

학위논문(석사) - 한국정보통신대학교 : 공학부, 2009.2, [ vii, 53 p. ]

Keywords

관심 영역; 비트 율 제어; 비디오 부호화기; 영상 통화; 확장된 관심 영역; Rate Control; ROI; xROI; Video encoder; Video telephony

URI
http://hdl.handle.net/10203/55060
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=393081&flag=t
Appears in Collection
School of Engineering-Theses_Master(공학부 석사논문)
Files in This Item
There are no files associated with this item.
  • Hit : 298
  • Download : 0
  • Cited 0 times in thomson ci

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0