Reference Picture List Constraints for Subpicture Random Access
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards face challenges in efficiently managing subpicture types and their decoding order, output order, and prediction relationships, particularly in multi-layer video coding contexts, which can lead to inefficiencies and increased bandwidth demand.
Innovation Solution
Implementing specific format rules for video processing that govern the conversion between video and bitstream, including constraints on subpicture types, decoding and output orders, and reference picture lists to ensure proper ordering and reduce unnecessary dependencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If video coding standards allow mixing of different subpicture types within a picture, then the versatility and adaptability of the coding system is improved, but the clarity of NAL unit type definitions and decoding processes deteriorates
Solution Approach 1:
The patent segments the picture into distinct subpicture regions, each with clearly defined NAL unit types (IDR, CRA, RADL, RASL). By partitioning the picture and assigning specific subpicture types to specific regions, the patent maintains versatility in supporting multiple subpicture types while preventing confusion through clear spatial segmentation and explicit type definitions for each subpicture region.
2Reliability
If clear constraints are imposed on subpicture types and their relationships, then the reliability of decoding is improved, but the complexity of the coding system increases
Solution Approach 1:
The patent applies local quality by imposing specific constraints on subpicture types in different regions of the picture. Each subpicture region has locally defined type requirements (e.g., leading subpictures must be RADL or RASL, trailing subpictures must be TRAIL). This localized constraint approach ensures decoding reliability without requiring complex global constraints throughout the entire picture.
Solution Approach 2:
Instead of allowing any subpicture type anywhere and adding complex restrictions, the patent inverts the approach by defining what each subpicture type MUST contain and where it MUST appear. For example, it defines that a leading subpicture MUST be of type RADL or RASL, and a trailing subpicture MUST be of type TRAIL. This inversion simplifies the constraint system by establishing positive requirements rather than negative restrictions.
3Adaptability or versatility
If the number of VCL NAL unit types within a picture is increased, then the adaptability of the coding system is improved, but the difficulty of detecting and measuring subpicture types increases
Solution Approach 1:
The patent uses distinct NAL unit type identifiers (analogous to color changes) to clearly differentiate between subpicture types. Each subpicture type (IDR, CRA, RADL, RASL, TRAIL) is associated with specific NAL unit type values, making it easy to detect and identify subpicture types by examining the NAL unit type field. This explicit typing system eliminates ambiguity even when multiple subpicture types coexist in a single picture.
Data Source
AI summary
Methods and apparatus for processing of video are described. The processing may include video encoding, decoding, or transcoding. One example video processing method includes performing a conversion between a video including one or more pictures including one or more subpictures and a bitstream of the video. The bitstream conforms to a format rule that specifies that a subpicture cannot be a random access type of subpicture in response to the subpicture not being a leading subpicture of an intra random access point subpicture. The leading subpicture precedes the intra random access point subpicture in output order.