Merge Mode for Template-based Intra Mode Derivation (TIMD) in ECM

March 24, 2025 Β· Declared Dead Β· πŸ› IEEE International Conference on Multimedia and Expo

πŸ‘» CAUSE OF DEATH: Ghosted
No code link whatsoever

"No code URL or promise found in abstract"

Evidence collected by the PWNC Scanner

Authors Mohsen Abdoli, Ramin G. Youvalari, Frank Plowman, Alexandre Tissier arXiv ID 2503.18679 Category cs.MM: Multimedia Citations 0 Venue IEEE International Conference on Multimedia and Expo Last Checked 4 months ago
Abstract
This paper presents an intra coding tool, named Merge mode for Template-based Intra Mode Derivation (TIMD). TIMD-Merge has been adopted in the 15\textsuperscript{th} version of the Enhanced Compression Model (ECM) software that explores video coding technologies beyond Versatile Video Coding (VVC) standard. This proposed tool operates on top of the regular TIMD mode that applies a template-based search on a causal adjacent template at top and at left of the current block, in order to find the best Intra Prediction Modes (IPMs) that matches the template. The proposed TIMD-Merge in this paper addresses a shortcoming in the regular TIMD method where due to texture discrepancy, the adjacent template information around the block is not reliable. To do so, the proposed TIMD-Merge constructs a list of all TIMD-coded blocks in relatively larger template area than the template of the regular TIMD, which also includes non-adjacent neighboring blocks. This list, called the merge list, is then sorted on the template to give one best set of TIMD modes. The use of TIMD-Merge mode is signalled at the block level and the implementation in the ECM-14.0 demonstrates -0.08\% performance improvement in terms of luma BDR gain, with negligible encoding and decoding runtime increase of 100.6% and 100.2%, respectively.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

πŸ“œ Similar Papers

In the same crypt β€” Multimedia

R.I.P. πŸ‘» Ghosted

Video Generation From Text

Yitong Li, Martin Renqiang Min, ... (+3 more)

cs.MM πŸ› AAAI πŸ“š 300 cites 8 years ago

Died the same way β€” πŸ‘» Ghosted