Video Encoding Asymmetric Block Partitioning for Motion Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding methods using fixed block sizes, such as 4×4 or 8×8, are inefficient for predicting and compensating for various video motions like zoom, shake, and rotation, leading to decreased encoding efficiency due to their inability to adapt to changing video characteristics.
Innovation Solution
A video encoding apparatus and method that partitions blocks into asymmetric shapes and sizes, using a scale factor to determine optimal partition forms and predict blocks, generating residual blocks, transforming, quantizing, and encoding these for improved compression efficiency and picture quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed block sizes (4×4 or 8×8) are used for partitioning, then the encoding process is simple, but the prediction accuracy for various video motions (zoom, shake, rotation) deteriorates
Solution Approach 1:
The current block is divided into multiple partition blocks with different asymmetric partition forms (e.g., 1:3, 3:1, 1:2, 2:1 ratios). This segmentation allows the encoding system to adapt to various motion patterns by selecting the most appropriate partition form for each block, thereby improving prediction accuracy without requiring a complete overhaul of the encoding structure.
Solution Approach 2:
The patent introduces dynamic block partitioning where the partition form is selected adaptively based on motion characteristics. Instead of using fixed block sizes, the system dynamically chooses from multiple asymmetric partition forms (1:3, 3:1, 1:2, 2:1) to match the actual motion patterns in the video content, enabling the encoding process to respond to varying motion complexity throughout the video.
2Productivity
If fixed block sizes are used, then the encoding method is simple and fast, but the adaptability to different video characteristics deteriorates
Solution Approach 1:
Different regions of the video content are processed with different partition forms based on their local motion characteristics. The system evaluates motion patterns in each block and applies the most suitable asymmetric partition form (1:3 for horizontal motion, 3:1 for vertical motion, etc.), ensuring that each local region is encoded with the optimal partition structure for its specific characteristics.
Solution Approach 2:
The patent changes the partitioning parameters dynamically by selecting from multiple asymmetric partition forms (1:3, 3:1, 1:2, 2:1) based on motion characteristics. This parameter adaptation allows the encoding system to adjust its behavior to match different video content types and motion patterns while maintaining a relatively simple encoding framework.
3Measurement precision
If various asymmetric partition forms are used, then the prediction accuracy for different motions is improved, but the device complexity increases
Solution Approach 1:
The system implements dynamic selection among a limited set of asymmetric partition forms (1:3, 3:1, 1:2, 2:1) based on motion characteristics. This dynamic approach allows the encoder to adapt to different motion patterns without requiring an excessive number of partition options, thereby balancing prediction accuracy with manageable device complexity.
Solution Approach 2:
Each asymmetric partition form is optimized for specific motion patterns (horizontal, vertical, or diagonal). By applying the appropriate partition form locally based on motion type, the system achieves high prediction accuracy for various motions while keeping the overall encoding structure relatively simple through targeted rather than universal complexity.
Data Source
AI summary
A video encoding apparatus, includes: a block partitioning unit to determine a partition form, among candidate partition forms, for partitioning a current block into one or more partition blocks; a prediction unit to generate one or more predicted blocks respectively corresponding to the one or more partition blocks by using a scale factor which indicates a ratio of pixel sampling; a subtractor to generate a residual block of the current block by subtracting the predicted blocks from one or more partition blocks corresponding thereto; a transformer to generate at least one transform block by transforming the residual block; a quantizer to generate at least one quantized transform block by quantizing the at least one transform block; and an encoder to encode, into the bitstream, information on the scale factor, information on the determined partition form and the at least one quantized transform block.