|Fast HEVC to VVC SCC transcoder by configurable decision tree based intra-frame coding
|Chan, Yui-lam (EIE)
Hong Kong Polytechnic University -- Dissertations
|Department of Electronic and Information Engineering
|30 pages : color illustrations
|SCC is proposed for encoding screen video content due to the special characteristics of desktop content. Combined with existing video coding standards such as HEVC and VCC, SCC can further improve the coding efficiency of desktop content. VVC SCC has better coding efficiency than HEVC SCC, but VVC SCC has higher coding complexity and takes a longer time than HEVC SCC. For existing desktop video contents which have already been encoded by HEVC, we can transcode them to VVC SCC to save the storing space and transmitting bandwidth. This thesis proposes a configurable and flexible fast HEVC to VVC SCC transcoding algorithm (FHVST), which aims to satisfy the needs of various HEVC to VVC SCC transcoding application scenarios. FHVST utilizes the features from both the HEVC decoder side and VVC encoder side to reduce the mode decision time during the VVC intra-frame coding process. In decoder side, two features are extracted from the training sequences, while in encoder side, fourteen features are extracted, and all these sixteen features from all training sequences are used to make the decision trees by machine learning. These decision trees are integrated into VVC SCC encoder for fast determining the coding modes(Intra, IBC, PLT) of CUs during the VVC SCC encoding process. In this thesis, CU in VVC partitioning is classified into four categories, which are defined as MR-CU (Major Rectangular CU), N4NSS-CU (Non-4xN Sized Square CU), 4NS-CU (4xN Sized CU), and Other CUs. The experimental results show that FHVST can bring 16.5% coding time reduction on average and 1.21% BDBR(Bjøntegaard delta bit rate) increase on average for MR-CU, and reduce 11% coding time with 1.20% BDBR increase for N4NSS-CU, and save 21% coding time with 2.58% BDBR increase for 4NS-CU for the YUV444 screen content sequences. When FHVST is applied to both MR-CU and N4NSS-CU at the same time, it can save 28% coding time and bring 3.79% BDBR increase on average. When FHVST applies to MR-CU, N4NSS-CU and 4NS-CU at the same time, it can save 51.8% coding time over average, but it brings 8.6% BDBR increase on average in this case.
|All rights reserved
Files in This Item:
|For All Users (off-campus access for PolyU Staff & Students only)
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: