Mode-Based Reference Filtering for Low-Power Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding methods do not efficiently utilize filters in inter prediction, leading to suboptimal resource usage and increased power consumption.
Innovation Solution
Adaptively applying filters to reference samples based on the inter prediction mode of a current block during inter prediction, optimizing memory bandwidth and reducing power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If filters are applied to reference samples in inter prediction, then prediction accuracy is improved, but memory bandwidth usage and power consumption increase
Solution Approach 1:
The patent applies filtering adaptively based on the inter prediction mode of the current block. Different filtering operations are dynamically selected depending on whether the block uses planar, DC, or other prediction modes, allowing the system to optimize between prediction accuracy and power consumption on a per-block basis rather than applying filters uniformly to all blocks
Solution Approach 2:
The patent applies filtering selectively to specific reference samples based on their location and the prediction mode being used. For example, planar prediction mode applies filtering to reference samples in a specific pattern, while DC mode applies different filtering, allowing local optimization of prediction quality where it matters most while reducing unnecessary filtering operations
2Measurement precision
If filters are applied to reference samples in inter prediction, then prediction accuracy is improved, but resource efficiency decreases
Solution Approach 1:
The system dynamically adjusts filtering operations based on the prediction mode of each block, enabling resource-efficient adaptive filtering that optimizes prediction accuracy only where necessary for different block types and prediction modes
Solution Approach 2:
Different filtering operations are applied locally to reference samples based on their specific prediction mode requirements, ensuring that filtering resources are allocated efficiently to where they provide the most benefit for prediction accuracy
Data Source
AI summary
Provided is a video decoding method including: determining an inter prediction mode of a current block when the current block is inter-predicted; determining at least one reference sample location to be referred to by the current block, based on the inter prediction mode of the current block; determining filter information to be applied to at least one reconstructed reference sample corresponding to the at least one reference sample location, based on the inter prediction mode of the current block; performing filtering on the at least one reconstructed reference sample, based on the filter information; and decoding the current block by using prediction samples generated via the filtering.