Video Stabilization via Feature Point Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video stabilization for fixed location cameras is challenging due to shaking caused by wind, vibrations, and moving objects, which can lead to dizziness and reduced effectiveness of video analytics, and existing methods struggle to differentiate between camera motion and object movement, resulting in strong artifacts.
Innovation Solution
A method and system that identify salient feature points, drop points associated with advancing and shaking objects, compute transformations between frames based on non-dropped points, and apply a stabilizing transformation to each frame to achieve stabilization, reducing processing by only stabilizing frames when displayed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all feature points are used for transformation computation, then more data is available for stabilization, but artifacts from moving objects increase
Solution Approach 1:
The patent extracts and removes feature points associated with moving objects from the set of feature points used for transformation computation. By identifying and eliminating these problematic feature points before computing the transformation, the method prevents artifacts from moving objects while maintaining reliable stabilization based on stationary feature points.
2Reliability
If stabilization is applied to all frames, then video quality is improved, but processing time and computational resources increase
Solution Approach 1:
The patent applies stabilization selectively rather than continuously to all frames. By determining which frames require stabilization based on display requirements and applying transformations only to those frames, the method maintains video quality where needed while reducing unnecessary processing time and computational resource consumption.
3Quantity of substance
If feature points from shaking objects are included, then more transformation data is obtained, but stabilization precision deteriorates
Solution Approach 1:
The patent extracts and removes feature points associated with shaking objects from the transformation computation. By eliminating these low-quality feature points that would degrade stabilization precision, the method maintains high accuracy while still utilizing the abundant feature points from stationary objects in the scene.
Data Source
AI summary
A method, system, and computer program product for stabilizing frames, the method comprising: receiving a frame sequence comprising three or more frames, including a current frame; determining salient feature points within the frames; matching the salient feature points between the frames; dropping salient feature points associated with advancing objects; dropping salient feature points associated with objects moving in shaking movements; computing a transformation between pairs of consecutive frames from amongst the at least three frames, based upon non-dropped salient feature points, thereby obtaining a multiplicity of transformations; determining a center position for the frames based upon the multiplicity of transformations; determining a stabilizing transformation from a current frame to the center position; and applying the stabilizing transformation to the current frame to obtain a stabilized frame.