Motion Vector Resolution Scaling for Screen Content Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding/decoding technologies do not efficiently account for the unique characteristics of screen content, such as clear pixel-based object boundaries and low noise, leading to suboptimal encoding/decoding efficiency.
Innovation Solution
The method and apparatus set motion vector resolutions to integer-pixel units based on screen content characteristics, incorporating flag information in header information to indicate integer-pixel resolutions, and perform motion prediction and encoding/decoding on a per-integer-pixel basis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion vector resolutions are set to sub-pixel units for natural images, then motion prediction accuracy is improved, but encoding complexity and bit rate increase
Solution Approach 1:
The patent applies different motion vector resolutions to different content types within the same video stream. Screen content regions use integer-pixel resolution while natural image regions use sub-pixel resolution, allowing each region to have the appropriate quality level for its characteristics
Solution Approach 2:
The patent changes the motion vector resolution parameter based on content type detection. When screen content is detected, the system switches from sub-pixel to integer-pixel resolution, adapting the parameter to match the content characteristics and avoid unnecessary complexity
2Productivity
If motion vector resolutions are set to integer-pixel units for screen content, then encoding efficiency is improved, but motion prediction accuracy for natural images deteriorates
Solution Approach 1:
The patent segments the video content into screen content regions and natural image regions, applying different motion vector resolutions to each segment. This allows integer-pixel resolution to be used for screen content (improving encoding efficiency) while sub-pixel resolution is used for natural images (maintaining accuracy)
Solution Approach 2:
The patent makes the motion vector resolution dynamic rather than fixed, allowing it to change based on the detected content type. The system dynamically switches between integer-pixel and sub-pixel resolutions frame-by-frame or region-by-region, optimizing performance for each specific content type
3Measurement precision
If sub-pixel motion vectors are used for all blocks, then motion prediction precision is improved, but the number of bits for encoding increases
Solution Approach 1:
The patent applies higher precision (sub-pixel) motion vectors only where necessary (natural image regions) and uses lower precision (integer-pixel) where sufficient (screen content regions), optimizing the balance between accuracy and bit rate
Solution Approach 2:
Instead of applying sub-pixel precision universally (excessive action), the patent applies it partially only to regions where it provides actual benefit, avoiding the unnecessary bit rate increase that would result from universal application
Data Source
AI summary
A video encoding method according to an embodiment of the present invention includes generating header information that includes information about resolutions of motion vectors of respective blocks, determined based on motion prediction for a unit image. Here, the header information includes flag information indicating whether resolutions of all motion vectors included in the unit image are integer-pixel resolutions. Further, a video decoding method according to another embodiment of the present invention includes extracting information about resolutions of motion vectors of each unit image from header information included in a target bitstream to be decoded; and a decoding unit for decoding the unit image based on the resolution information. Here, the header information includes flag information indicating whether resolutions of all motion vectors included in the unit image are integer-pixel resolutions.