Video Coding Apparatus Motion Vector Prediction Composite Reference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies, such as MPEG-4 AVC, face challenges in maintaining prediction accuracy and reducing information transmission due to degradation in reference image quality, especially under high compression and non-uniform motion conditions, leading to increased computation and coding costs.
Innovation Solution
A video coding apparatus that detects motion vectors between a coding target block and a reference image, composites this with other reference images to create a prediction image, and codes the prediction difference block, thereby reducing prediction residuals and improving image quality without increasing motion vector information transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If motion compensated prediction is used with high compression coding, then compression ratio is improved, but prediction accuracy deteriorates due to degradation component mixed in the reference image
Solution Approach 1:
The patent extracts the degradation component from the reference image by comparing multiple reference images and identifying pixels that differ significantly. This extracted degradation information is then used to correct the prediction image, effectively removing the harmful compression artifacts while maintaining high compression ratios.
Solution Approach 2:
The patent composites multiple reference images to create a corrected prediction image. By combining information from multiple reference images and selectively applying degradation correction based on pixel differences, the system creates a composite prediction that maintains accuracy while achieving high compression.
2Loss of information
If motion vector prediction is used assuming temporal continuity, then amount of information transmission is reduced, but prediction accuracy deteriorates when motion is not temporally continuous
Solution Approach 1:
The patent introduces a feedback mechanism where the system detects actual motion by comparing multiple reference images and uses this detected motion information to correct prediction errors. This feedback loop allows the system to adapt to non-uniform motion patterns while maintaining efficient information transmission through selective transmission of correction data.
3Measurement precision
If two reference images are used for motion compensated prediction, then prediction image quality is improved by smoothing degradation, but amount of coding increases due to additional motion vectors
Solution Approach 1:
The patent extracts only the essential degradation correction information from multiple reference images rather than transmitting complete motion vectors for each. By taking out only the necessary correction data and applying it selectively, the system improves prediction quality while avoiding the coding overhead of multiple full motion vectors.
4Measurement precision
If motion compensated prediction is used with small blocks, then prediction accuracy is improved by capturing object boundary motion, but amount of coding increases due to more motion vectors
Solution Approach 1:
The patent merges the benefits of small-block prediction accuracy with the efficiency of larger block coding by applying degradation correction at the pixel level across the entire prediction image. This allows the system to capture fine motion details at object boundaries without requiring separate motion vectors for each small block, thus maintaining accuracy while reducing coding complexity.
Data Source
AI summary
A motion vector detector detects a motion vector from a first reference image for a coding target block. A multiple reference image composition unit creates a composite reference block that a first reference block extracted from the first reference image using the motion vector has been composited with a predetermined region of at least one of other reference images. Setting the composite reference block as a prediction block, an entropy encoder encodes a prediction difference block that the prediction block has been subtracted from the coding target block.