Motion Vector Offset Ranking for Efficient Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in improving coding efficiency, enhancing image quality, reducing processing amounts and circuit scale, and optimizing selection of elements like filters and motion vectors.
Innovation Solution
The proposed method involves determining an initial motion vector and generating an ordered list of candidate offsets based on similarity with nearby pixels, selecting an offset for the initial motion vector, and employing apparatuses for video encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional motion vector determination methods are used, then processing amount and circuit scale are reduced, but coding efficiency and image quality deteriorate
Solution Approach 1:
The patent segments the motion vector determination process into distinct stages: initial motion vector calculation, candidate offset generation, cost calculation, and offset selection. This segmentation allows the system to achieve high coding efficiency through multiple refinement stages while managing circuit complexity by organizing functionality into modular units that can be implemented efficiently.
Solution Approach 2:
The patent performs preliminary action by calculating an initial motion vector before generating candidate offsets. This preliminary motion vector serves as a foundation for subsequent refinement steps, allowing the system to achieve high coding efficiency by starting from a reasonable approximation rather than searching from scratch, thereby reducing the overall processing burden.
2Manufacturing precision
If conventional motion vector determination methods are used, then processing amount is reduced, but image quality and coding efficiency deteriorate
Solution Approach 1:
The patent employs dynamic adaptation by generating candidate offsets based on local pixel characteristics and motion patterns. The system dynamically adjusts the candidate offset set according to the specific block being processed, using cost calculations that adapt to local image content. This dynamic approach achieves high image quality by tailoring the motion vector search to local characteristics while managing processing complexity through selective refinement.
Solution Approach 2:
The patent changes parameters by introducing multiple candidate offsets with different precision levels and selecting the optimal offset based on cost calculations. The system varies the granularity and number of candidate offsets depending on the block characteristics, achieving high manufacturing precision (image quality) by using fine-grained offsets when needed while reducing processing amount by using coarser offsets for simpler blocks.
3Productivity
If multiple candidate offsets are evaluated, then coding efficiency and image quality improve, but processing amount increases
Solution Approach 1:
The patent applies partial action by evaluating multiple candidate offsets but not exhaustively searching all possible offsets. The system generates a limited set of promising candidate offsets based on initial motion vector and local characteristics, then evaluates only these candidates using cost calculations. This approach achieves high coding efficiency by focusing computational effort on the most promising candidates rather than performing exhaustive searches, thereby reducing processing amount and energy consumption.
Solution Approach 2:
The patent uses feedback mechanisms by calculating costs for each candidate offset and using this cost information to select the optimal offset. The cost calculation provides feedback on the quality of each candidate, allowing the system to efficiently identify the best offset without evaluating all possible options. This feedback-driven selection achieves high coding efficiency while managing processing amount by eliminating poor candidates early in the evaluation process.
Data Source
Figure 1
Figure 2(a)~2(f)
Figure 3~4
AI summary
The present disclosure relates to methods and apparatuses for motion vector, MV, determination that may be used in video encoding and/or decoding. The motion vector determination obtains an initial MV associated with a current block and generates an ordered list of candidate offsets for the current block. The generating includes obtaining a set of offsets relative to the initial MV, and for each offset in the set of offsets (1) determining a cost reflecting a similarity between an offset MV obtained involving combination of the initial MV and said offset, and one or more predetermined MVs associated with pixels in proximity of the current block; (2) deciding whether said offset is added into the ordered list of candidate offsets, and (3) in case it is decided that said offset is added into the ordered list, decide at which position said offset is to be added to the ordered list based on the cost. The motion vector determination then determines a selected offset for said initial MV from the ordered list of the candidate offsets.