Video Encoder B-Frame Reference Segmentation for Complexity Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video encoding using bi-predictively coded frames (B-frames) incurs high computational complexity due to the use of multiple reference frames, which existing technologies have not adequately addressed.
Innovation Solution
Implementing techniques such as not using B-frames as reference frames, simplifying cost functions for motion estimation, and performing fractional pixel refinement only in one direction to reduce computational complexity, including switching off the de-blocking filter for B-frames and using a lower complexity filter for motion search.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If B-frames are used as reference frames for encoding P-frames and other B-frames, then compression efficiency is improved, but computational complexity increases
Solution Approach 1:
The patent segments the reference frame usage by frame type: B-frames are used as reference frames for I-frames and P-frames, but not for other B-frames. This segmentation allows the encoder to benefit from B-frame reference data for inter-frame prediction while avoiding the computational complexity of using B-frames as references for subsequent B-frames, thus resolving the contradiction between compression efficiency and computational complexity
Solution Approach 2:
The patent applies partial action by selectively disabling de-blocking filtering only for B-frames while maintaining it for I-frames and P-frames. This partial application reduces computational complexity specifically where B-frame characteristics allow it, without compromising overall visual quality, thus addressing the contradiction between processing complexity and output quality
2Manufacturing precision
If de-blocking filter is applied to B-frames, then visual quality is improved, but computational complexity and power consumption increase
Solution Approach 1:
The patent applies de-blocking filtering selectively: it is applied to I-frames and P-frames but disabled for B-frames. This partial application maintains visual quality where it matters most (for frames that serve as references for future frames) while reducing computational complexity and power consumption for B-frames, thus resolving the contradiction between visual quality and computational complexity
3Measurement precision
If fractional pixel refinement is performed for both past and future reference frames, then motion estimation accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies different levels of motion estimation refinement to different reference frames based on their local characteristics. Fractional pixel refinement is performed selectively on the reference frame that provides better motion compensation, rather than uniformly on both past and future reference frames. This local optimization maintains motion estimation accuracy while reducing unnecessary computational complexity
Solution Approach 2:
The patent performs fractional pixel refinement partially - only on one reference frame (either past or future) rather than both. This partial application provides sufficient motion estimation accuracy for B-frame encoding while significantly reducing the computational complexity associated with performing refinement on both reference frames
Data Source
AI summary
Several techniques aimed at reducing computational complexity when encoding uses bi-predictively encoded frames (B-frames) are implemented in a video encoder. In an embodiment, B-frames are not used as reference frames for encoding P-frames and other B-frames. Non-use of B-frames allows a de-blocking filter used in the video encoder to be switched off when reconstructing encoded B-frames, and use of a lower complexity filter for fractional-resolution motion search for B-frames. In another embodiment, cost functions used in motion estimation for B-frames are simplified to reduce computational complexity. In one more embodiment, fractional pixel refinement in motion search for B-frames is simplified. In yet another embodiment, predictors used in motion estimation for a macro-block in a P-frame are selected from a B-frame that uses a same reference frame as the P-frame.