Screen Video Coding Prediction Using Block Type Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing screen video coding technologies face challenges in efficiently compressing and predicting screen video content, particularly with high-frequency data such as text, lines, and graphics, which often result in blurring or disappearance during the coding process.
Innovation Solution
The method involves classifying coding blocks into different types, such as text and natural images, and using a computing device to filter candidate blocks and compute motion vector sets for type-based motion merge and advanced motion vector prediction modes, thereby improving inter/intra prediction efficiency by filtering out search points with different block types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional motion merge mode or AMVP mode is used to select motion vectors from multiple candidates, then coding flexibility is improved, but computing complexity increases due to comparing multiple motion vectors and selecting the best match
Solution Approach 1:
The patent applies preliminary action by pre-classifying coding blocks into different types (text, natural image, mixed) before motion vector prediction. This classification is performed once and reused for multiple motion vector candidates, avoiding repeated classification computations during the motion merge and AMVP processes, thus reducing computing complexity while maintaining coding flexibility
Solution Approach 2:
The patent applies local quality by using different block type classification results for different regions of the screen video. Each coding block is classified according to its local characteristics (text-heavy, natural image-heavy, or mixed), and these local classification results are used to guide motion vector selection in corresponding regions, improving coding efficiency without requiring global analysis
2Manufacturing precision
If screen video coding uses high-frequency data representation for text and graphics, then image quality is improved, but blurring or disappearance occurs during compression due to high-frequency component loss
Solution Approach 1:
The patent applies parameter changes by adjusting coding parameters based on block type classification. For text blocks, it uses different prediction modes and motion vector selection strategies compared to natural image blocks. This adaptive parameter adjustment maintains high-frequency details in text and graphics regions while applying appropriate compression to natural image regions, preventing blurring and disappearance
Solution Approach 2:
The patent applies local quality by applying different coding strategies to different block types within the same video frame. Text blocks receive specialized handling with preserved high-frequency components, while natural image blocks use conventional compression, ensuring that text and graphics remain sharp and visible while maintaining overall video quality
3Productivity
If classifier is used to classify coding blocks into different types, then coding efficiency is improved, but device complexity increases due to additional classification computations
Solution Approach 1:
The patent applies preliminary action by performing block type classification once before motion vector prediction and reuseing these classification results for multiple subsequent operations. The classifier processes each coding block once to determine its type (text, natural image, or mixed), and this classification information is then used for both motion merge mode and AMVP mode, avoiding repeated classification computations and reducing overall device complexity
Solution Approach 2:
The patent applies universality by designing the block type classification mechanism to serve multiple functions simultaneously. The same classification results are used for both motion merge mode candidate selection and AMVP mode search point filtering, making the classification system multi-functional and reducing the need for separate processing mechanisms
Data Source
AI summary
According to one to one exemplary embodiment, the disclosure provides a method of coding prediction for screen video. The method classifies a plurality of coding blocks into a plurality of block types by using a classifier; and uses a computing device to filter at least one candidate block from the plurality of coding blocks, according to the plurality of block types of the plurality of coding blocks, and compute a first candidate motion vector set of a type-based motion merge mode and a second candidate motion vector set of a type-based advanced motion vector prediction mode, wherein each of the at least one candidate block has a block-type different from that of a current coding block.