Neural Network In-Loop Filter Padding for Adaptive Slice Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding techniques struggle with accommodating the varying sizes and types of video slices and temporal layers in multilayer bitstreams, leading to inefficiencies in bandwidth usage and coding processes.
Innovation Solution
The method determines real-time padding dimensions for video units to adjust the padding area, considering factors such as location, size, color format, slice type, and neural network filtering methods, and performs conversions between video units and bitstreams based on these dimensions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional fixed padding sizes are used for all video units, then the coding process is simple, but the bandwidth utilization is inefficient and coding performance deteriorates for different slice types and temporal layers
Solution Approach 1:
The padding dimensions are made dynamic and adaptive rather than fixed. The system determines padding dimensions in real-time based on video unit characteristics including slice type, temporal layer, color format, and size. This allows the padding area to be optimized for each specific video unit context, improving coding efficiency without requiring complex manual configuration
Solution Approach 2:
Different padding dimensions are applied to different video units based on their specific characteristics. Instead of using a uniform padding size across all video units, the system tailors the padding dimensions to match the local requirements of each video unit, slice type, and temporal layer, thereby optimizing coding performance for diverse video content
2Loss of energy
If padding dimensions are optimized for different slice types and temporal layers, then bandwidth utilization improves, but the real-time determination complexity increases
Solution Approach 1:
The system performs preliminary determination of padding dimensions by evaluating video unit characteristics such as slice type, temporal layer, color format, and size before the actual encoding process. This advance determination allows the optimized padding dimensions to be applied during encoding, improving bandwidth utilization without adding significant real-time complexity during the critical encoding path
Data Source
AI summary
A method implemented by a video coding apparatus. The method includes determining, in real time, padding dimensions for padding samples to be applied to a video unit of a video for in-loop filtering, wherein d1, d2, d3, and d4 represent the padding dimensions corresponding to top, bottom, left, and right boundaries of the video unit, respectively; and performing a conversion between a video unit and a bitstream of the video based on the padding dimensions that were determined. A corresponding video coding apparatus and non-transitory computer-readable recording medium are also disclosed.