Parameterized Motion Models for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional video compression schemes rely on regular motion compensation, which assumes translational motion and fails to accurately predict non-translational motions like scaling, shearing, or rotating motions within video frames, especially when multiple global motions are present.
Innovation Solution
The use of parameterized motion models, such as homographic, affine, similarity, and translational models, to generate prediction blocks for encoding and decoding video frames, allowing for more accurate prediction of various types of motion by selecting the appropriate model based on lowest prediction error.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional regular motion compensation is used, then device complexity is reduced, but manufacturing precision of motion prediction deteriorates
Solution Approach 1:
The patent applies parameter changes by transitioning from simple translational motion parameters to parameterized motion models that include scaling, shearing, and rotating parameters. This allows the system to adapt the complexity of motion representation to match the actual motion characteristics in the video content, improving prediction accuracy while managing coding complexity through selective application.
Solution Approach 2:
The patent implements dynamics by making the motion model adaptable and selectable based on content requirements. Different parameterized motion models can be dynamically chosen for different blocks or regions, allowing the system to optimize between simplicity and accuracy depending on the specific motion patterns present in each region of the video frame.
2Manufacturing precision
If parameterized motion models are used, then motion prediction accuracy is improved, but data requirements increase
Solution Approach 1:
The patent applies local quality by applying parameterized motion models selectively to specific blocks or regions where complex motion patterns are detected, rather than uniformly across the entire frame. This allows high prediction accuracy where needed while maintaining lower data rates in regions with simple translational motion, optimizing the balance between quality and bit rate.
3Adaptability or versatility
If multiple parameterized motion models are supported, then adaptability is improved, but device complexity increases
Solution Approach 1:
The patent applies partial action by implementing a hierarchical or progressive approach to motion model selection, where simpler models are attempted first and more complex models are only invoked when necessary. This reduces the average computational burden while maintaining the adaptability to use sophisticated models when the video content requires them.
Data Source
AI summary
A current block of a video frame can be encoded or decoded using parameterized motion models. First and second parameterized motion models are identified. The first parameterized motion model corresponds to a first motion model type, and the second parameterized motion model corresponds to a second motion model type. The first and second parameterized motion models are associated with one or more reference frames. One of the first or second parameterized motion models is selected along with an associated reference frame, such as based on a lowest prediction error. A motion vector is generated between the current block and the selected reference frame by warping pixels of the current block to a warped patch of the selected reference frame according to the selected parameterized motion model. A prediction block is generated using the motion vector, and the current block is encoded or decoded using the prediction block.