Decoded Picture Buffer State Signaling for Robust Reference Lists
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video codec standards face challenges in efficiently managing and updating reference picture buffers and lists, leading to complications and robustness issues due to loss of state-affecting information, which can adversely affect decoding performance.
Innovation Solution
Explicitly signaling state information about decoded picture buffers and reference picture lists using flags and decision flags, such as future reference flags, on/off decision flags, and higher layer picture designation, to simplify management and improve robustness against information loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If elaborate rules for managing and updating reference pictures are implemented, then compression efficiency is improved, but decoder complexity and robustness against information loss deteriorate
Solution Approach 1:
The patent extracts the reference picture management state information from the implicit decoder internal state and makes it explicit by signaling it in the bitstream. This separates the complex management logic from the decoder implementation, allowing the decoder to simply read the signaled state rather than compute it, thereby reducing decoder complexity while maintaining compression efficiency.
Solution Approach 2:
The patent introduces feedback by having the encoder signal the current state of the decoded picture buffer and reference picture lists to the decoder. This feedback mechanism ensures the decoder can synchronize with the encoder's reference picture selection without implementing complex management rules, improving robustness against information loss while maintaining compression efficiency.
2Ease of operation
If internal state tracking is used for DPB and RPL management, then decoding operations can be performed, but robustness against information loss deteriorates
Solution Approach 1:
The patent creates a copy of the encoder's reference picture buffer state information and transmits it to the decoder through the bitstream. This copied state information allows the decoder to reconstruct the reference picture lists without relying on its own internal state tracking, which may be corrupted by information loss, thereby improving robustness while maintaining ease of decoding operations.
Solution Approach 2:
The patent performs preliminary action by having the encoder pre-compute and signal the reference picture buffer state before the decoder needs to use it. This advance provision of state information ensures the decoder has accurate reference picture lists even if subsequent bitstream information is lost, enhancing robustness without complicating the decoding process.
3Device complexity
If reference picture management is simplified, then decoder complexity is reduced, but compression efficiency deteriorates
Solution Approach 1:
The patent introduces an intermediary mechanism where the signaled buffer description list acts as a mediator between the encoder's reference picture management and the decoder's decoding process. This intermediary carries the necessary state information to maintain compression efficiency while allowing the decoder to use simple read operations instead of complex management logic, thus reducing decoder complexity without sacrificing compression efficiency.
Data Source
AI summary
Innovations for signaling state of a decoded picture buffer (“DPB”) and reference picture lists (“RPLs”). In example implementations, rather than rely on internal state of a decoder to manage and update DPB and RPLs, state information about the DPB and RPLs is explicitly signaled. This permits a decoder to determine which pictures are expected to be available for reference from the signaled state information. For example, an encoder determines state information that identifies which pictures are available for use as reference pictures (optionally considering feedback information from a decoder about which pictures are available). The encoder sets syntax elements that represent the state information. In doing so, the encoder sets identifying information for a long-term reference picture (“LTRP”), where the identifying information is a value of picture order count least significant bits for the LTRP. The encoder then outputs the syntax elements as part of a bitstream.