Block-Level Reference Picture Reordering for Low-Overhead Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques lack block-level adaptations in reference picture lists for inter prediction, leading to increased signaling overhead that outweighs coding gains.
Innovation Solution
Implementing an adaptive reference picture reordering process at the block level to generate new orderings of reference pictures without additional signaling overhead, reducing the bits needed for inter prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If block-level reference picture list adaptation is implemented, then coding gain is improved, but signaling overhead increases
Solution Approach 1:
The decoder autonomously reorders reference pictures in the reference picture list based on block-level characteristics without requiring explicit signaling from the encoder. The system uses locally available information (motion vectors, block characteristics) to determine the optimal reference picture ordering, making the adaptation self-sufficient and avoiding additional bitstream overhead
Solution Approach 2:
The reference picture list ordering is dynamically changed at block level based on local video characteristics such as motion vector magnitude and block type. Different blocks may have different reference picture orderings optimized for their specific characteristics, allowing coding gain improvement without uniform signaling overhead across the entire picture
2Adaptability or versatility
If additional signaling is added for block-level adaptation, then reference picture list adaptation is enabled, but signaling overhead outweighs coding gains
Solution Approach 1:
The adaptation mechanism operates autonomously at the decoder side using locally available block characteristics. The decoder determines the reference picture ordering based on motion information and block properties already present in the bitstream, eliminating the need for additional signaling to convey adaptation parameters
Solution Approach 2:
The reference picture list is prepared and reordered in advance during the decoding process based on predicted optimal ordering derived from block characteristics. This preliminary organization enables efficient inter prediction without requiring real-time signaling during the decoding of each block
Data Source
AI summary
A video decoder may be configured to generate a first ordering of reference pictures in a reference picture list for a first block of a slice, wherein generating the first ordering of the reference pictures for the first block comprises assigning indexes to the reference pictures; and generate a second ordering of the reference pictures in the reference picture list for a second block of the slice based on an adaptive reference picture reordering process, wherein the first ordering is different than the second ordering and generating the second ordering of reference pictures for the second block comprises assigning at least some of the indexes to different reference pictures than in the first ordering; decode the first block using the first ordering of the reference pictures; and decode the second block using the second ordering of the reference pictures.


