| The existing video coding standard H.264although in recent years has made greatachievement, but as the video quality requirements of high and ultrahigh definition videocontents in multimedia applications, and the storage and transmission ability is limited by thebandwidth limits, so the video compression ability requires more and more high, H.264already can’t meet the requirements. The video industry experts VCEG organization andMPEG organization joint form a new group JCT-VC, this organization dedicates toformulate a new generation of video coding standard HEVC. Although HEVC adds manynew technology to improve the coding efficiency, but it does not consider human visualsystem. JND is one of the most main models, this paper proposes to join JND characteristicsto the HEVC framework, in order to guarantee the subjective video quality and reducecoding bit rate.In this paper, we first introduce the new video coding standard HEVC, by presentingthe key technology now used in the coding frame. HEVC focuses on the study of a newcoding tools or technology, but still belong to mixed encoding frame as H.264. Andcompared to H.264, the macro-block size increases to64×64, which is more advantageousto the high resolution video compression, and will be divided into three levels of: codingunits (CU), prediction units (PU) and transform unit (TU). In the respect of the Humanvisual system, this paper first introduces of two kinds of main JND models: pixel domainJND model and DCT domain model. At last, the JND model is separately applied in theHEVC coding guiding the residual filter in pixel domain and the DCT coefficients filter inDCT domain The results show that the original HEVC coding compare to the coding framewhich introduce the JND model in pixel domain can enhance the coding efficiency oncondition that they have the almost same subjective quality, and in DCT domain，the DCTcoefficients JND filter does not work well. |
PDF Full Text Request