Image Stabilization Vignetting Removal and Reintroduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image stabilization in videos often results in vignetting, which can be annoying and amplifies noise, especially when different image frames are shifted and cropped to compensate for camera shake.
Innovation Solution
A method that involves determining a dislocation value for each digital image frame, applying a vignetting removal process, and reintroducing a uniform vignetting effect based on the data defining the vignetting effect and dislocation value, minimizing noise and flickering in the stabilized video.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If image stabilization is performed by shifting and cropping image frames, then camera shake is reduced and video stability is improved, but vignetting becomes apparent and moves across frames causing annoyance
Solution Approach 1:
The harmful vignetting effect is extracted and removed from the image frames through a vignetting removal process. This separates the unwanted vignetting from the stabilized image content, allowing the video to remain stable without the annoying moving vignetting artifacts.
Solution Approach 2:
The vignetting removal is applied selectively to specific regions of the image frames. By determining a dislocation value for each frame and applying vignetting removal based on the dislocation, the solution targets only the affected areas while preserving the rest of the image quality.
2Object-generated harmful factors
If vignetting removal is applied to all pixels, then vignetting is eliminated, but noise is heavily amplified in the affected areas
Solution Approach 1:
Instead of applying vignetting removal uniformly to all pixels, the method applies it locally only to pixels corresponding to the determined region based on the dislocation value. This selective application removes vignetting where needed while avoiding noise amplification in unaffected areas.
Solution Approach 2:
The vignetting removal process is applied partially rather than completely to all image data. By removing vignetting only from the necessary regions and reintroducing it elsewhere, the solution achieves sufficient vignetting correction without the excessive noise amplification that would result from full-frame processing.
3Object-generated harmful factors
If vignetting removal process is applied differently to each frame, then moving vignetting is eliminated, but processing complexity and time increase
Solution Approach 1:
The vignetting effect is pre-characterized by providing data that defines how different pixels are affected by vignetting. This preliminary characterization allows subsequent frames to be processed more efficiently by referencing the pre-defined vignetting model rather than analyzing each frame from scratch.
Solution Approach 2:
The method uses the dislocation value determined for each frame as feedback to control the vignetting removal and reintroduction process. This feedback mechanism allows the system to adapt to frame-specific conditions while maintaining consistency with the overall vignetting model, reducing unnecessary processing complexity.
Data Source
AI summary
The present invention relates to a method for enabling an image stabilized video. The method comprises providing data defining a vignetting effect for digital image frames captured by a video camera; determining a dislocation value for a digital image frame captured by the video camera; determining, based on the dislocation value, a region of the digital image frame to be displayed in the image stabilized video; applying a vignetting removal process at least on pixels of the digital image frame corresponding to the region to be displayed in the image stabilized video, wherein said applying being based on the data defining the vignetting effect; and reintroducing vignetting on the region of the digital image frame to be displayed in the video based on the data defining the vignetting effect and the dislocation value.