Reference Picture Management in Library-Based Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding methods limit the number of reference pictures that can be used, restricting coding gain and inefficiently managing dependency relations between video sequences and library pictures, particularly in library-based video coding.
Innovation Solution
The method specifies reference pictures by extracting signaling flags from a reference mapping table to indicate different numeration principles, allowing the decoder to correctly identify and obtain reference pictures from library layers, even if they do not belong to the same temporal segment, and processes requests for reference pictures to ensure accurate decoding and efficient transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional video coding methods are used to limit the number of reference pictures, then device complexity is reduced, but coding gain is restricted
Solution Approach 1:
The patent segments the reference picture management into two distinct layers: video layer bitstreams and library layer bitstreams. This segmentation allows independent management of reference pictures for video content and library content, enabling the system to utilize multiple reference pictures from different layers without increasing overall system complexity. The reference mapping table is also segmented to map reference picture identifiers from both layers, facilitating efficient reference picture selection and reducing the burden on single-layer reference management.
2Productivity
If library-based video coding is implemented to improve coding gain, then dependency relation management becomes more complex
Solution Approach 1:
The patent introduces a reference mapping table as an intermediary structure between video layer bitstreams and library layer bitstreams. This table stores and manages the dependency relations between reference pictures from different layers, acting as a mediator that simplifies the complex many-to-many relationships. The reference mapping table enables the decoder to efficiently resolve which reference pictures from the library layer should be used for predicting current video pictures, without requiring complex real-time calculations or management logic.
Solution Approach 2:
The patent performs preliminary action by pre-establishing the reference mapping table that defines all dependency relations between video and library layer bitstreams before decoding occurs. The reference picture identifiers and their corresponding mappings are predetermined and stored, allowing the decoder to simply look up and use the appropriate reference pictures without performing complex runtime analysis of dependency relations. This preliminary setup significantly reduces the computational complexity during actual video decoding.
3Measurement precision
If multiple numeration principles are used for reference picture identification, then reference picture accuracy is improved, but decoding complexity increases
Solution Approach 1:
The patent applies different numeration principles locally to different types of reference pictures based on their source layer. Video layer reference pictures use one numeration principle (e.g., sequential numbering), while library layer reference pictures use another numeration principle (e.g., hierarchical or content-based numbering). The reference mapping table explicitly indicates which numeration principle applies to each reference picture identifier, allowing the decoder to select the appropriate interpretation method for each reference picture without requiring a single complex numeration system for all references.
Data Source
AI summary
Methods to produce media data ensure a method for media data transmitting, and a method for media data processing, and the synchronization and correct processing and transmission of bitstreams on the basis of dependency relation between video layer bitstream and library layer bitstream, and correct bitstream is highly efficiently provided to a decoder. With methods for reference picture request processing and reference picture specifying, a current picture includes a library picture set in which the library picture does not belong to the random access segment to which the current picture belongs and the previously most adjacent random access segment, and correct decoding of the current picture is ensured and repeated downloading of library pictures is avoided, thus ensuring the correct decoding and highly efficient transmission of a bitstream produced by library-based video coding method, and increasing transmission efficiency and storage efficiency.