Intra-prediction Virtual Boundaries Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face challenges in efficiently performing intra-prediction across virtual boundaries in video data, particularly during gradual decoder refresh, where blocks on the clean side of a picture cannot be predicted from blocks on the dirty side, leading to increased bitrate and processing complexity.
Innovation Solution
The implementation of techniques that code data representing virtual boundaries within a picture, disabling the use of blocks on the dirty side for intra-prediction references of blocks on the clean side, and using pulse code modulation or intra-prediction from clean regions of previously coded pictures, allowing for efficient encoding and decoding without relying on dirty side samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If blocks on the clean side are predicted using blocks on the dirty side, then prediction accuracy is improved, but bitrate and processing complexity increase
Solution Approach 1:
The picture is segmented into a clean side and a dirty side using a virtual boundary. Blocks on the clean side are predicted using only reference samples from the clean side, while blocks on the dirty side may use samples from both sides. This segmentation prevents the need to track and manage references across the boundary, reducing processing complexity while maintaining prediction accuracy within each region.
Solution Approach 2:
The harmful factor of using dirty side samples for clean side prediction is extracted and eliminated by disabling such predictions through the virtual boundary mechanism. This removes the complexity of managing cross-boundary references while preserving the beneficial use of clean side samples for both regions.
2Measurement precision
If blocks on the clean side are predicted using blocks on the dirty side, then prediction accuracy is improved, but bitrate increases
Solution Approach 1:
By segmenting the picture into clean and dirty sides with a virtual boundary, the patent enables more efficient prediction for clean side blocks using only clean side references. This reduces the number of bits needed to signal prediction information and prevents the bitrate increase that would result from managing cross-boundary predictions.
3Productivity
If virtual boundaries are implemented for gradual decoder refresh, then decoding efficiency is improved, but device complexity increases
Solution Approach 1:
The virtual boundary segments the picture into regions with different prediction rules, enabling gradual decoder refresh where the clean side can be decoded independently of the dirty side. This segmentation improves decoding efficiency by allowing parallel processing and reducing dependency chains, while the boundary management adds minimal complexity compared to the overall decoding process.
Solution Approach 2:
The virtual boundary and clean/dirty side classification are established in advance before prediction is performed. This preliminary action allows the decoder to quickly determine which reference samples are valid without complex runtime decisions, improving decoding efficiency while keeping the added complexity manageable through pre-computed boundary information.
Data Source
AI summary
An example device for coding (encoding or decoding) video data includes a memory configured to store video data; and one or more processors implemented in circuitry and configured to: code data representing a location of a virtual boundary defining a clean side of a picture of video data and a dirty side of the picture; disable use of blocks on the dirty side for intra-prediction reference of blocks on the clean side; and code the picture without using the blocks on the dirty side for intra-prediction reference of the blocks on the clean side.