Local Global Prediction Modes for Non-Translational Motion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques are limited in efficiently representing non-translational motion, leading to inefficient bandwidth utilization and inaccurate compression, particularly when using translational motion models and inefficient signaling of warped motion parameters.
Innovation Solution
The use of local global prediction modes with projected motion fields, where warped motion model parameters are signaled at the superblock level, reducing resource utilization by temporal propagation of the motion field, and encoding and decoding processes are optimized to handle motion parameters effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If translational motion models are used for video coding, then the coding process is simple, but the accuracy of representing non-translational motion deteriorates
Solution Approach 1:
The video frame is divided into multiple superblocks, and each superblock is further divided into blocks. Different motion models (translational, affine, perspective) are applied to different superblocks based on their motion characteristics. This segmentation allows the system to use simple translational models where appropriate while applying more accurate affine or perspective models where non-translational motion is detected, thus balancing complexity and accuracy.
Solution Approach 2:
The patent implements dynamic motion model selection where the coding system adaptively chooses between translational, affine, and perspective motion models based on the actual motion characteristics of each superblock. The system calculates motion parameters for each model and selects the one that provides the best prediction accuracy, allowing the complexity to vary dynamically according to the content being encoded rather than using a fixed complex model throughout.
2Loss of energy
If warped motion parameters are signaled efficiently, then bandwidth utilization improves, but the complexity of handling multiple motion models increases
Solution Approach 1:
The patent performs preliminary classification of superblocks into different motion model categories before encoding. The system evaluates motion characteristics and determines which superblocks require affine or perspective models versus those that can use simpler translational models. This preliminary action allows the decoder to be prepared with the appropriate motion model information, reducing the need for complex real-time decisions during decoding and enabling more efficient bandwidth utilization.
Solution Approach 2:
The patent introduces an intermediate motion parameter representation layer that bridges the gap between complex warped motion models and efficient bandwidth utilization. By deriving compact motion parameters from affine and perspective models and signaling them in a standardized format, the system acts as an intermediary that preserves motion accuracy while reducing the bandwidth overhead associated with transmitting full warped motion parameter sets.
3Measurement precision
If affine and perspective motion models are used, then non-translational motion representation accuracy improves, but bandwidth requirements increase
Solution Approach 1:
The patent applies different motion model complexities to different local regions (superblocks) of the video frame based on their specific motion characteristics. Rather than using affine or perspective models globally, the system identifies superblocks exhibiting non-translational motion patterns and applies the appropriate complex model only to those regions. This local quality approach ensures high accuracy where needed while maintaining bandwidth efficiency in regions where simpler models suffice.
Solution Approach 2:
The patent dynamically changes the parameters of the motion models based on the content being encoded. For superblocks with simple translational motion, the system uses fewer parameters (translational model). For superblocks with rotational or scaling motion, it transitions to affine models with additional parameters. For complex perspective transformations, it uses perspective models with even more parameters. This parameter adaptation allows the system to optimize the balance between accuracy and bandwidth by matching parameter complexity to motion complexity.
Data Source
AI summary
Coding using local global prediction modes with projected motion fields includes identifying a current frame, identifying a current reference frame, obtaining a projected motion field, for the current frame, using motion data from the current reference frame, identifying a current superblock from the current frame, obtaining reference warp motion parameters for the current superblock by fitting the projected motion field to a warp motion model, and using the reference warp motion parameters to code respective blocks from the superblock.