Inter Prediction Using Segmented Picture Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inter prediction methods in video encoding and decoding are complex and inefficient, especially when dealing with large picture blocks, leading to high hardware resource consumption.
Innovation Solution
The proposed method involves determining a plurality of first picture blocks within a to-be-processed picture block based on preset split widths and heights, and performing bi-directional optical flow prediction on these blocks to obtain predictors. The predictors are then combined to obtain a predictor for the to-be-processed picture block, thereby reducing the complexity of inter prediction and improving processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If inter prediction is performed on large picture blocks, then processing coverage is improved, but complexity of implementing inter prediction increases
Solution Approach 1:
The patent divides a large picture block into multiple smaller sub-blocks (first picture blocks) based on preset split widths and heights. Bi-directional optical flow prediction is then performed separately on each sub-block, which reduces the computational complexity compared to processing the entire large block as a single unit, while still maintaining comprehensive coverage of the original picture block.
2Productivity
If picture block size is increased, then compression efficiency is improved, but memory resource consumption increases
Solution Approach 1:
By segmenting large picture blocks into smaller sub-blocks, the patent reduces the amount of data that needs to be held in memory simultaneously for each processing unit. This segmentation allows the system to process large areas effectively while consuming less memory resources at any given time, as each sub-block requires fewer memory resources than the entire large block.
Solution Approach 2:
The patent applies different processing approaches to different regions by performing localized bi-directional optical flow prediction on each sub-block. This local processing approach optimizes memory usage by focusing computational resources on smaller regions at a time, rather than loading and processing the entire large picture block in memory simultaneously.
3Productivity
If picture block size is increased, then processing speed is improved, but hardware resource consumption increases
Solution Approach 1:
The patent segments large picture blocks into smaller sub-blocks for parallel processing. This segmentation enables the system to achieve high processing speeds by processing multiple sub-blocks concurrently or in efficient sequences, while reducing hardware resource consumption compared to processing one large block at a time, as the computational load is distributed across multiple smaller units.
Data Source
AI summary
This application discloses an inter prediction method and apparatus. The method includes: determining a plurality of first picture blocks in a to-be-processed picture block based on a preset picture split width, a preset picture split height, and a width and a height of the to-be-processed picture block; separately performing bi-directional optical flow prediction on the plurality of first picture blocks to obtain a predictor of each first picture block; and obtaining a predictor of the to-be-processed picture block with a combination of the predictors of the plurality of first picture blocks. The apparatus includes a determining module, a prediction module, and a combination module. In this application, complexity of implementing inter prediction can be reduced, and processing efficiency can be improved.