Long-term Reference Picture Signaling Simplification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The HEVC long-term picture signaling mechanism is overly complex, requiring five lists to be generated and numerous restrictions to be tracked, which increases the complexity of decoder design and power consumption.
Innovation Solution
A method for decoding a current picture from a bitstream that simplifies long-term picture handling by decoding code words in a loop to derive positive integers for each long-term picture, comparing indicator values with stored pictures, and marking unused long-term pictures for removal from the decoded picture buffer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the HEVC long-term picture signaling mechanism is used to ensure accurate reference picture management, then picture decoding reliability is improved, but decoder complexity and power consumption increase
Solution Approach 1:
The patent segments the reference picture management into two distinct parts: short-term reference pictures (ST-RPS) and long-term reference pictures (LT-RPS). Each type is handled with its own simplified signaling mechanism. ST-RPS uses a compact representation with offset values, while LT-RPS uses a separate list with explicit POC values. This segmentation allows the decoder to process each type independently with reduced complexity compared to the unified HEVC approach.
Solution Approach 2:
The patent extracts the long-term reference picture management from the complex HEVC RPS structure and creates a separate, simplified LT-RPS mechanism. By taking out the LT-RPS handling from the main RPS processing flow, the decoder can manage long-term pictures with a dedicated, less complex algorithm that doesn't require tracking multiple interconnected lists and restrictions.
2Measurement precision
If the HEVC long-term picture signaling mechanism with five lists and numerous restrictions is implemented, then reference picture identification accuracy is improved, but power consumption increases
Solution Approach 1:
The patent employs a more compact and disposable representation for long-term reference pictures. Instead of maintaining five complex lists with numerous restrictions, the LT-RPS uses a simple structure with explicit POC values and a single used_by_curr_pic_lt_flag per picture. This simplified structure requires less computational resources to process and maintain, directly reducing power consumption while preserving identification accuracy.
3Reliability
If five lists are generated and numerous restrictions tracked in HEVC, then long-term picture reference accuracy is improved, but the number of processing operations increases
Solution Approach 1:
The patent merges the long-term reference picture information into a dedicated LT-RPS structure that is processed separately from short-term pictures. This combining of LT-RPS handling into a unified, simplified flow eliminates the need to track multiple restrictions across five different lists. The decoder processes LT-RPS with a single coherent algorithm, improving decoding efficiency while maintaining reference accuracy.
Data Source
AI summary
A method decodes one or more code words in a loop that is run N times by decoding an indication of whether a picture is a long-term picture or not and deriving a positive integer K[i]>0 for each long-term picture. An indicator value V[i] is derived for each long-term picture by subtracting the value of the positive integer K[i] from a picture indicator value P of the current picture. For each of the indicator values V[i], the indicator value V[i] is compared with values associated with pictures stored in the DPB. The method keeps the pictures stored in the DPB marked as short-term or long-term for which there is a match with one of the indicator values V[i] in the DPB as long-term pictures, and marks long-term pictures for which there is no match with any indicator value V[i] as unused for reference in the DPB.