Video Slice and Tile Signaling for Efficient Picture Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video coding standards, such as HEVC and VVC, lack efficient methods for signaling and parsing picture partition information, particularly in scenarios involving multiple tiles and slices, which can lead to inefficiencies in processing and decoding.
Innovation Solution
A method and apparatus for signaling picture partition information by modifying the syntax to support partitioning a picture into slices and tiles, including signaling the number of tiles and CTU rows within a slice, and determining the position and shape of slices, ensuring efficient decoding and in-loop filtering across tile boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If picture partition information is not efficiently signaled, then decoding complexity increases, but compression efficiency deteriorates
Solution Approach 1:
The picture is divided into multiple tiles and slices, with each partition having independent signaling. The syntax structure separates tile partition information (num_tiles_in_pic, tile_width_in_ctbs, tile_height_in_ctbs) from slice partition information (num_slices, slice_height_in_ctbs), allowing incremental decoding where each partition can be processed independently without requiring the entire picture to be decoded first.
2Measurement precision
If detailed partition information is signaled for every slice, then decoding precision improves, but bitstream overhead increases
Solution Approach 1:
The signaling approach adapts to local picture characteristics by allowing different partition configurations for different regions. Each slice can have its own height in CTBs (slice_height_in_ctbs), and tiles can have different dimensions (tile_width_in_ctbs, tile_height在ctbs), enabling optimized partitioning for local content requirements without uniformly over-signaling the entire picture.
Solution Approach 2:
Instead of signaling the starting position of each slice from the top of the picture, the patent signals the height of each slice (slice_height_in_ctbs). This inverted approach allows slices to be defined by their dimensions rather than positions, reducing the number of bits required while maintaining the same partitioning information.
3Reliability
If in-loop filtering is applied across tile boundaries, then picture quality improves, but processing complexity increases
Solution Approach 1:
The filter across tiles enabled flag (filter_across_tiles_enabled_flag) provides dynamic control over filtering behavior. When set to true, in-loop filtering is applied across tile boundaries; when false, filtering is restricted within tiles. This dynamic switching allows the system to adapt to different content types and quality requirements, enabling high-quality filtering when needed while maintaining lower complexity for simpler content.
Data Source
AI summary
A method and apparatus for signaling or parsing picture partition information are disclosed. According to one method, a syntax related to a number of tiles in a target slice is signaled at an encoder side or parsed at a decoder side if the address of the target slice does not correspond to a last tile index of the current picture and the raster-scan slice mode is used for picture partition. The syntax is signaled or parsed from a slice header level of the video bitstream corresponding to the target slice. According to another method, a control syntax is signaled at an encoder side or parsed at a decoder side only if the picture partition information indicates more than one tile existing in the current picture. The control syntax indicates whether in-loop filtering is applied across tile boundaries.