Video Frame Stabilization Using Subject POV and 3D Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video stabilization methods fail to effectively remove unwanted camera motion and noise from video sequences captured in unstable conditions, leading to jitter and visual disturbances.
Innovation Solution
A method and electronic device that determine the point of view of a subject in a video sequence, segment frames into foreground and background segments, and stabilize the sequence by aligning and transforming objects based on 3D structure fusion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional video stabilization methods are used to remove unwanted camera motion, then camera jitter and visual disturbances are reduced, but the method fails to effectively remove anomalous motions and distortions such as camera shifts, scaling, and rotation
Solution Approach 1:
The patent segments the video frame into multiple regions (first region containing the subject and second region containing background) and applies different stabilization processing to each region. This allows the system to handle anomalous motions in the subject region separately from background motions, improving the removal of camera shifts, scaling, and rotation while maintaining natural appearance.
Solution Approach 2:
The patent applies different stabilization characteristics to different regions of the video frame. The first region (subject) receives stabilization processing tailored to remove anomalous motions, while the second region (background) receives appropriate processing to maintain natural appearance. This local differentiation enables effective removal of harmful factors without compromising overall video quality.
2Reliability
If video stabilization processing is applied to remove camera motion, then unwanted shaking and jitter are reduced, but the complexity of the processing system increases
Solution Approach 1:
The patent divides the stabilization processing into distinct modules: region segmentation module, first stabilization processing module for the subject region, and second stabilization processing module for the background region. This modular segmentation reduces overall system complexity by allowing independent optimization of each processing stage.
Solution Approach 2:
The patent performs preliminary region segmentation before applying stabilization processing. By pre-dividing the frame into subject and background regions, the system can apply appropriate stabilization techniques to each region without needing to process the entire frame uniformly, thereby reducing computational complexity while maintaining effectiveness.
Data Source
AI summary
A method for stabilization of a video sequence captured by an electronic device is provided. The method includes identifying a subject in the video sequence, estimating a velocity of the subject relative to the electronic device, determining a point of view of a subject in the video sequence with respect to the electronic device and the velocity of the subject relative to the electronic device and stabilizing the video sequence based on the determined point of view.