Video Tile Slice Index Mapping for Decoder Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards face challenges in accurately signaling slice and subpicture information, leading to incorrect derivation of CTU counts and potential decoder crashes, especially when rectangular slices are split across tiles and subpictures.
Innovation Solution
The proposed solution involves deriving picture-level and subpicture-level slice indices and creating a mapping function between these indices to correctly determine the number of CTUs in each slice, ensuring consistent decoding order and accurate parsing of slice data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If slice partitioning information is included in syntax structure of video unit, then accurate slice identification is improved, but bitstream complexity increases
Solution Approach 1:
The patent derives the number of CTUs in each slice in advance during encoding, and stores this information in the bitstream. During decoding, the decoder directly uses this pre-calculated information without needing to perform complex runtime calculations, thus improving slice identification accuracy while avoiding bitstream complexity
Solution Approach 2:
The patent introduces an intermediary variable (number of CTUs per slice) that acts as a bridge between the slice partitioning structure and the decoding process. This intermediary stores the essential information needed for accurate slice identification without requiring the decoder to reconstruct the entire partitioning logic, thereby simplifying the bitstream structure while maintaining precision
2Adaptability or versatility
If rectangular slices are split across tiles and subpictures, then picture coverage flexibility is improved, but decoder reliability deteriorates
Solution Approach 1:
The patent segments the picture into tiles and subpictures, and further segments each slice into CTUs. By deriving and storing the number of CTUs in each slice, the patent enables flexible slicing across tile and subpicture boundaries while maintaining decoder reliability through explicit counting information that prevents out-of-bounds access and ensures complete picture coverage
Solution Approach 2:
The patent uses feedback by deriving the number of CTUs based on the slice partitioning configuration and storing this information in the bitstream. The decoder reads this feedback information to verify correct slice parsing and to prevent decoder crashes, thus maintaining reliability even when slices are split across multiple tiles and subpictures
3Adaptability or versatility
If CTU count derivation is performed dynamically, then decoding adaptability is improved, but processing time increases
Solution Approach 1:
The patent performs the computationally intensive CTU count derivation during the encoding phase and stores the results in the bitstream. During decoding, the decoder simply reads the pre-calculated values without performing complex calculations, thus maintaining decoding adaptability while significantly reducing processing time
Solution Approach 2:
The patent calculates and stores more information than strictly minimal (the complete CTU count for each slice) during encoding. This excessive action during encoding allows the decoder to perform only simple reading and verification operations, trading encoding complexity for decoding speed and reducing overall processing time
Data Source
AI summary
A method includes performing a conversion between a video including a video picture including one or more tiles and a bitstream of the video. The video picture refers to a picture parameter set, and the picture parameter set conforms to a format rule specifying that the picture parameter set includes a list of column widths for N tile columns, where N is an integer. An (N−1)-th tile column exists in the video picture and the (N−1)-th tile column has a width that is equal to an (N−1)-th entry in a list of explicitly included tile column widths plus one number of coding tree blocks.


