Video Stabilization via Homography Matrix Curve Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video stabilization methods, particularly those using optical and sensor-based approaches, are costly, inflexible, and computationally complex, while electronic methods for image processing are time-consuming and limited in application scenarios, especially in addressing the rolling shutter effect.
Innovation Solution
A method and system for video stabilization that divides target frames into pixel groups, determines offset corrections using curve fitting, and applies correction homography matrices to stabilize videos captured with rolling shutters, reducing computational complexity and enhancing flexibility across various scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If optical video stabilization or sensor-based electronic stabilization is used, then video stabilization effect is improved, but device cost and complexity increase
Solution Approach 1:
The patent replaces optical stabilization mechanisms (lens position adjustment) and sensor-based stabilization (gyroscope hardware) with a computational approach using homography matrices and curve fitting algorithms. This substitutes mechanical and hardware-based solutions with software-based image processing, achieving stabilization without additional physical components.
Solution Approach 2:
The patent creates a corrected copy of the video frames through homography transformation. Instead of physically stabilizing the camera or using sensors to prevent vibration, the method generates a stabilized version of the captured frames by applying geometric transformations, effectively copying and correcting the visual content.
2Device complexity
If electronic video stabilization via image processing is used, then device cost is reduced, but processing time and computational complexity increase
Solution Approach 1:
The patent divides the image processing task into segmented steps: detecting feature points in each frame, computing homography matrices based on these points, and applying corrections frame-by-frame. This segmentation allows for optimized processing of individual frames rather than treating the entire video as one complex task, reducing overall processing time.
Solution Approach 2:
The patent computes homography matrices using only selected feature points rather than all pixels in the frame. By identifying and processing only the necessary feature points that capture the essential motion information, the method achieves stabilization with reduced computational effort compared to full-frame processing.
3Device complexity
If traditional electronic video stabilization methods are used, then implementation is simplified, but adaptability to different scenarios is limited
Solution Approach 1:
The patent employs dynamic homography matrices that are computed for each frame based on detected feature points and motion characteristics. This allows the stabilization parameters to adapt dynamically to different shooting scenarios, camera motions, and video content, rather than using fixed stabilization settings.
Solution Approach 2:
The patent changes key parameters (homography matrix values, curve fitting coefficients) based on the specific characteristics of each video frame and shooting scenario. By adjusting these parameters dynamically according to the detected motion and feature point distribution, the method achieves high adaptability across different video stabilization scenarios.
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
A method for video stabilization may include obtaining a target frame of a video; dividing a plurality of pixels of the target frame into a plurality of pixel groups; determining a plurality of first feature points in the target frame; determining first location information of the plurality of first feature points in the target frame; determining second location information of the plurality of first feature points in a frame prior to the target frame in the video; obtaining a global homography matrix; determining an offset of each of the plurality of first feature points; determining a fitting result based on the first location information and the offsets; for each of the plurality of pixel groups, determining a correction matrix; and for each of the plurality of pixel groups, processing the pixels in the pixel group based on the global homography matrix and the correction matrix.