Video Coding Motion Vector Selection Using Adjoining Block Reference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing coding methods, such as H.26x and HEVC, face inefficiencies in coding non-flat images with high noise conditions due to incorrect detection of minute motion vectors and the inability to select merge or skip modes when motion vectors of adjoining blocks are not determined, leading to reduced coding efficiency.
Innovation Solution
A coding apparatus that detects motion vectors for each block, calculates difference values between motion vectors and pixel differences, and selects the appropriate motion vector for coding based on threshold values, allowing for efficient coding even when the motion vector of an adjoining block has not been determined.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion search is performed for each block independently, then motion vector detection can be completed, but in high noise conditions minute motion vectors are incorrectly detected and merge/skip modes cannot be selected when adjoining block motion vectors are not yet determined
Solution Approach 1:
The patent performs motion search for the left adjoining block before processing the target block, ensuring the motion vector is determined in advance. This preliminary action allows the merge mode to be properly evaluated and selected when appropriate, even in high noise conditions, by having the necessary motion vector information available before target block processing begins.
Solution Approach 2:
The patent uses the motion vector of the left adjoining block as a reference for the target block when performing motion search. By copying or referencing the already-determined motion vector from the adjoining block, the system can more reliably determine whether to use merge mode, reducing incorrect detections in high noise conditions while maintaining coding efficiency.
2Productivity
If the motion vector of the left adjoining block is used for merge mode selection, then coding efficiency can be improved, but in pipeline processing the motion vector is not determined at the time of motion search for the target block
Solution Approach 1:
The patent restructures the pipeline processing to perform motion search for the left adjoining block before initiating motion search for the target block. This preliminary completion of adjoining block processing ensures that its motion vector is determined and available when needed for merge mode evaluation, without significantly increasing overall system complexity.
Solution Approach 2:
The patent introduces a control mechanism that coordinates between different processing stages, using the determined motion vector of the left adjoining block as an intermediary reference for target block processing. This mediator approach allows efficient use of available motion vector information while maintaining manageable pipeline processing complexity.
3Ease of manufacture
If conventional coding methods are used in high noise conditions, then standard processing can be maintained, but coding efficiency is reduced due to incorrect motion vector detection
Solution Approach 1:
The patent implements a feedback mechanism where the motion vector of the left adjoining block, which has already been determined through motion search, is fed back as a reference for processing the target block. This feedback allows the system to leverage previously obtained motion information, improving coding efficiency in high noise conditions while maintaining relatively simple implementation by building upon existing motion vector data.
Data Source
AI summary
A coding apparatus for coding a frame of a moving image on a block basis includes a calculation unit configured to, for a block adjoining the target block and having a motion vector to be used for coding having been selected, from among blocks in the frame, calculate a difference between a detected motion vector and a motion vector of the target block and calculate a difference between pixels of a reference block identified by applying a motion vector of the adjoining block to the target block and pixels of the target block, a determination unit configured to determine whether a motion vector to be used for coding the adjoining block has been detected at the time of processing, and a motion vector selection unit configured to, based on a calculation result and a determination result, select a motion vector to be used for coding the target block.