Screen Content Video Encoding Integer Motion Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding methods, particularly for screen-captured moving images, incur increased bandwidth and processing resources due to fractional pixel resolution for motion vectors, which is unnecessary since most motion vectors point to integer pixels, leading to potential quality loss and inefficiency.
Innovation Solution
An encoder with a motion prediction module that restricts motion vectors to integer pixel offsets for screen-captured content, while maintaining fractional precision for camera-captured video, reducing bit rate and processing complexity by switching between modes based on the source type of the moving image data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If fractional pixel resolution is used for motion vectors, then compression efficiency and quality are improved, but bit rate and processing resources increase
Solution Approach 1:
The patent applies different motion vector precision levels to different types of content: fractional pixel precision (1/4 pixel) is used for video content where smooth motion occurs, while integer pixel precision is used for screen content where motion is typically discrete. This local differentiation optimizes the balance between precision and bit rate for each content type.
Solution Approach 2:
The patent changes the precision parameter of motion vectors based on content type detection. By detecting whether the input is video or screen content and adjusting the motion vector precision accordingly, the system achieves optimal compression efficiency while minimizing unnecessary bit rate consumption.
2Measurement precision
If fractional pixel resolution is used for motion vectors, then matching accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent applies different motion estimation complexities to different content types: full fractional pixel search is performed for video content where it provides significant benefit, while integer pixel search is sufficient for screen content. This reduces processing complexity for screen content without sacrificing necessary matching accuracy.
Solution Approach 2:
The patent performs partial fractional pixel search only when necessary (for video content), rather than applying full fractional pixel precision to all content. This partial application of high-precision processing reduces overall computational burden while maintaining accuracy where needed.
3Quantity of substance
If integer pixel resolution is used for motion vectors in screen content, then bit rate is reduced, but quality may be degraded
Solution Approach 1:
The patent changes the motion vector precision parameter based on content type: integer pixel precision is used for screen content to reduce bit rate, while fractional pixel precision is used for video content to maintain quality. This dynamic parameter adjustment resolves the contradiction between bit rate and precision.
Solution Approach 2:
The patent applies appropriate precision levels locally to different content types: integer precision for screen content where it is sufficient, and fractional precision for video content where it provides quality benefits. This localized approach maintains encoding precision where needed while reducing bit rate where possible.
Data Source
Figure 1~3
Figure 4~5
Figure 6
AI summary
An input of an encoder receives moving image data comprising a sequence of frames to be encoded, each frame comprising a plurality of blocks in two dimensions with each block comprising a plurality of pixels in those two dimensions. A motion prediction module performs encoding by, for at least part of each of a plurality of said frames, coding each block relative to a respective reference portion of another frame of the sequence, with the respective reference portion being offset from the block by a respective motion vector. According to the present disclosure, the moving image data of this plurality of frames comprises a screen capture stream, and the motion prediction module is configured to restrict each of the motion vectors of the screen capture stream to an integer number of pixels in at least one of said dimensions.