Intra Block Copy Hash Classification for Screen Content Video
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards, such as HEVC, are not optimized for screen content video, which has different properties than natural camera-captured videos, leading to inefficiencies in compression and increased hardware implementation complexity due to the need for maintaining unfiltered reconstructed samples for intra block copy (IBC) operations.
Innovation Solution
The proposed solution involves a video coding method that uses intra block copy (IBC) mode to predict coding units within a picture by classifying blocks into categories based on hash values and determining an optimal block vector (BV) for displacement, allowing for efficient compression of screen content video while reducing implementation complexity by limiting reference samples to reconstructed blocks within the current and left neighboring CTUs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If intra block copy (IBC) mode is used to predict coding units within a picture, then coding efficiency is improved, but device complexity increases due to the need for maintaining unfiltered reconstructed samples
Solution Approach 1:
The current picture is divided into multiple non-overlapped blocks, and hash values are calculated for each block to identify potential reference blocks. This segmentation approach enables efficient IBC prediction by organizing reference samples in a structured manner, reducing the complexity of maintaining and searching unfiltered reconstructed samples across the entire picture.
Solution Approach 2:
The patent applies different processing to different regions by classifying blocks into categories based on their hash values. Reference samples are maintained locally within the current and left neighboring CTUs rather than globally, which reduces the overall hardware complexity while preserving coding efficiency in regions that benefit from IBC.
2Productivity
If reference samples are maintained across the entire picture for IBC operations, then coding performance is improved, but hardware complexity increases
Solution Approach 1:
The patent divides the picture into multiple non-overlapped blocks and maintains reference samples only within the current and left neighboring CTUs. This localized segmentation reduces the total number of reference samples that need to be maintained in hardware while still providing sufficient reference material for effective IBC prediction in screen content regions.
Solution Approach 2:
The patent changes the scope parameter of reference sample maintenance from global (entire picture) to local (current and left neighboring CTUs). This parameter change significantly reduces hardware complexity by limiting the reference sample buffer size while maintaining compression performance through strategic block classification and hash-based matching within the reduced scope.
Data Source
AI summary
A method for coding video data comprises receiving a video picture comprising a plurality of coding units. The picture is divided into multiple non-overlapped blocks. An encoder calculates a hash value of each block of the multiple non-overlapped blocks. All the non-overlapped blocks are classified into at least two categories comprising a first category and a second category. The first category comprises one or more blocks representing one or more hash values covered by a first set of hash values, and the second category comprises all remaining blocks. The blocks in the second category are classified into at least two groups including a first group. The first group comprises one or more blocks representing the same hash value as another block in the second category. An associated computing device and a non-transitory computer readable storage medium are also provided.