Screen Content Video Coding Search Scheme
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fast search methods for hybrid video encoders are not optimized for screen content, leading to inefficient inter-frame prediction and high computational complexity during desktop sharing, as they are designed based on camera video features rather than screen content features.
Innovation Solution
A direction-based search followed by a feature-oriented search is implemented to find a suitable predictor for inter-frame prediction, reducing computational complexity by comparing pixel blocks along predetermined directions and utilizing feature indices to determine similarity within a search window of a reference frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full search is used for inter-frame prediction, then prediction accuracy is improved, but computational complexity increases significantly
Solution Approach 1:
The patent segments the search process into two distinct phases: a fast search phase that provides an initial predictor, and a refinement phase that improves accuracy. This segmentation allows the system to avoid the computational burden of full search while still achieving good prediction results through targeted refinement of selected candidate blocks.
Solution Approach 2:
The patent applies local quality by using different search strategies for different regions and block types within the video content. The fast search uses simplified criteria suitable for general areas, while the refinement phase applies more rigorous evaluation only to promising candidate blocks, optimizing computational resources where they are most needed.
2Productivity
If conventional fast search methods are used for screen content, then computation speed is improved, but prediction accuracy deteriorates
Solution Approach 1:
The patent performs preliminary action by conducting a fast search to identify candidate predictor blocks before applying refinement. This preliminary fast search quickly narrows down the search space to promising candidates, ensuring computation speed is maintained while setting the stage for accuracy improvement in the subsequent refinement phase.
Solution Approach 2:
The patent implements feedback by using the results of the fast search to guide the refinement process. The fast search provides initial predictions that are evaluated, and this feedback information is used to select and refine the most promising candidate blocks, creating a closed-loop system that improves accuracy without sacrificing speed.
3Measurement precision
If full search is applied to screen content with large resolution, then prediction accuracy is improved, but encoding time increases excessively
Solution Approach 1:
The patent applies partial action by performing a complete fast search to identify candidate blocks, but only applying the more computationally intensive refinement process to a selected subset of these candidates rather than all of them. This partial application of refinement maintains prediction accuracy for the most important blocks while avoiding excessive encoding time that would result from refining all candidates.
Data Source
AI summary
A method, a device and computer readable storage media facilitate providing screen content including a plurality of video frames that are displayed by a computing device. During coding of the screen content, a suitable predictor is found that is used to code pixel blocks from one or more frames. The suitable predictor is found by selecting a pixel block of a current frame, conducting a direction based search by comparing pixel blocks within a search window of a reference frame with the selected pixel block of the current frame to determine whether a match exists, and, in response to a determination that no sufficient match has been found, conducting a feature oriented search by comparing pixel blocks of the reference frame with the selected pixel block of the current frame to find a suitable match based upon a common feature.