Segmented Motion Models for Affine Video Compression Bottlenecks
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 in data compression and decoding processes.
Innovation Solution
Implementing segmentation-based parameterized motion models that segment video frames into multiple segments with distinct motion models, allowing for more accurate prediction and encoding using parameterized motion vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If translational motion models are used for video compression, then the compression process is simple and fast, but the accuracy in representing complex motions (scaling, shearing, rotating) is poor
Solution Approach 1:
The patent divides the video frame into multiple motion segments, where each segment can have its own parameterized motion model (translational, rotational, scaling, shearing). This allows different regions of the frame to be represented with appropriate motion models, improving overall motion representation accuracy while keeping individual segment processing simple.
Solution Approach 2:
The patent introduces dynamic motion models that can adapt to different motion types (translational, rotational, scaling, shearing) based on the content of each motion segment. The motion model parameters are updated dynamically during encoding to best represent the actual motion characteristics of each segment.
2Measurement precision
If parameterized motion models with multiple segments are used, then motion representation accuracy is improved, but encoding and decoding complexity increases
Solution Approach 1:
By segmenting the frame into motion segments, the patent enables parallel processing of different segments during encoding and decoding. Each segment can be processed independently, which maintains accuracy while improving overall throughput through parallel computation.
Solution Approach 2:
The patent uses parameterized motion models with a fixed set of motion types (translational, rotational, scaling, shearing). This parameterization approach provides a balanced trade-off between representation accuracy and computational complexity, as the number of parameters is limited and can be efficiently processed.
3Measurement precision
If simple translational motion vectors are used, then data requirements are minimal, but decoding accuracy for complex motions is insufficient
Solution Approach 1:
The patent segments the video into motion segments, allowing different motion models to be applied to different segments. This segmentation enables the system to use simple translational vectors where appropriate while employing more complex parameterized models only where needed, optimizing the balance between data requirements and decoding accuracy.
Solution Approach 2:
The patent parameterizes motion models with a controlled set of motion parameters (translation, rotation, scaling, shearing). This parameterization limits the amount of data required while providing sufficient accuracy for complex motions, as the parameters are designed to capture essential motion characteristics without excessive complexity.
Data Source
AI summary
A current frame is segmented, at an encoder, with respect to a reference frame into multiple segments. Each segment represents different underlying motion. For each segment, a parameterized motion model is determined describing the underlying motion for blocks within that segment. For a block, a first prediction and a second prediction are evaluated, respectively, using the parameterized motion model and a translational motion vector. The parameterized motion model and an indication of which prediction to use are encoded into a bitstream. A decoder decodes a motion model type associated with a segment from a current frame header in a compressed bitstream. The motion model type is selected from similarity and affine motion model types. Parameters for a parameterized motion model are determined based on the decoded motion model type. A prediction block for a block is generated by applying a transformation defined by the determined parameters to a reference frame.