Intra Sub-Partition Coding Order for Lower Video Prediction Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face inefficiencies due to increased overhead in transmitting intra prediction modes for sub-split blocks, as the encoding order is fixed, preventing optimal use of closer reference pixels for prediction, especially in rectangular blocks.
Innovation Solution
Adaptive determination of the encoding/decoding order of subblocks based on their shape, size, sub-splitting direction, and prediction mode, allowing for flexible intra prediction using an ISP mode.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a block is further sub-split into smaller blocks for intra prediction, then prediction accuracy improves by using closer reference pixels, but the overhead for transmitting intra prediction modes for each sub-split block increases
Solution Approach 1:
The patent applies dynamics by making the encoding order adaptive rather than fixed. The encoding order is dynamically determined based on the prediction direction and block shape, allowing the system to flexibly adjust the processing sequence to optimize reference pixel selection while maintaining efficient mode transmission.
Solution Approach 2:
The patent changes the parameter of encoding order from a fixed value to an adaptive parameter that varies according to prediction direction and block characteristics. This parameter change enables the system to select optimal reference pixels without increasing mode transmission overhead.
2Ease of manufacture
If an fixed encoding order (left to right or top to bottom) is used for sub-split blocks, then the encoding process is simple, but reference pixels closer to the current subblock cannot be appropriately used when they are not located in the previous coding direction
Solution Approach 1:
The patent transforms the static encoding order into a dynamic one that adapts to prediction direction. By determining the encoding order based on prediction direction and block shape, the system maintains simplicity while enabling appropriate use of closer reference pixels.
Solution Approach 2:
The encoding order parameter is changed from a fixed sequence to an adaptive sequence that depends on prediction direction and block characteristics, allowing optimal reference pixel selection without complicating the overall encoding process.
3Loss of information
If one intra prediction mode is signaled for the current block to reduce overhead, then transmission efficiency improves, but intra prediction cannot be performed optimally on each sub-split block with different reference pixel requirements
Solution Approach 1:
The patent applies dynamics by making the reference pixel selection adaptive to each sub-split block's position and prediction direction, even when a single mode is transmitted. This allows optimal local prediction without increasing mode transmission overhead.
Solution Approach 2:
The patent applies local quality by allowing different reference pixel selections for different sub-split blocks based on their local characteristics (position, shape, prediction direction), while maintaining a single transmitted mode for the entire block to keep overhead low.
Data Source
AI summary
A video coding method and an apparatus using an adaptive order of intra sub-partitions are provided. The video coding method and apparatus adaptively determine an encoding/decoding order of subblocks based on at least one of a shape of the subblock, a size of the subblock, a sub-splitting direction, and a prediction mode in performing intra prediction of the subblocks split from a current block using an ISP mode.