Segmented Motion Models for Complex Video Frame Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques using translational motion models fail to accurately represent complex motions such as scaling, shearing, or rotating motions within video frames, leading to inefficiencies and increased data requirements.
Innovation Solution
Implementing segmentation-based parameterized motion models that segment video frames into multiple segments with associated motion models, allowing for more accurate prediction of block motions using parameterized motion models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If translational motion models are used for video compression, then encoding simplicity is maintained, but prediction accuracy for complex motions deteriorates
Solution Approach 1:
The video frame is divided into multiple segments, with each segment assigned its own motion model (translational, similarity, affine, or homographic). This segmentation allows different types of motion to be modeled appropriately in different regions, improving prediction accuracy while maintaining encoding efficiency through selective model application.
Solution Approach 2:
The patent employs parameterized motion models with varying degrees of complexity (translational with 2 parameters, similarity with 4 parameters, affine with 6 parameters, homographic with 8 parameters). By changing the number of parameters based on motion complexity requirements, the system achieves both encoding simplicity for simple motions and high prediction accuracy for complex motions.
2Measurement precision
If segmentation-based parameterized motion models are used, then prediction accuracy for complex motions is improved, but device complexity increases
Solution Approach 1:
The system dynamically selects the appropriate motion model type and parameters based on the actual motion characteristics of each segment. This dynamic adaptation allows the device complexity to scale with the required prediction accuracy, using simpler models when sufficient and more complex models only when necessary.
Solution Approach 2:
Different motion models with different complexities are applied to different segments of the video frame based on local motion characteristics. This local quality approach ensures that complex models are only used where needed, minimizing overall device complexity while maintaining high prediction accuracy for complex motions in specific regions.
3Productivity
If multiple motion models are used per frame, then encoding efficiency for complex motions is enhanced, but processing time increases
Solution Approach 1:
By segmenting the frame and assigning appropriate motion models to each segment, the system processes only the necessary portions with the required model complexity. This avoids applying complex models to entire frames when simpler models would suffice for most regions, thereby reducing overall processing time while maintaining encoding efficiency for complex motions.
Solution Approach 2:
The system adjusts the number of parameters based on motion complexity requirements for each segment. By using fewer parameters for simple motions and more parameters only where complex motions occur, the processing time is optimized while maintaining high encoding efficiency for the complex motion segments that require it.
Data Source
AI summary
Multiple global motion models associated with respective segments of a current frame are decoded from a compressed bitstream. Each global motion model is based on a segmentation of the current frame and represents a respective underlying motion of blocks within a respective segment. Blocks of the current frame are decoded by: for each inter-predicted block of a segment, decoding, form the compressed bitstream, an indication of whether to decode the each inter-predicted block based on a global motion model of the multiple global motion models and associated with the segment, or whether to decode the each inter-predicted block based on a motion vector that is different from the global motion model; and decoding the each inter-predicted block based on the indication.