Dynamic Multi-Exposure Camera Calibration for Artifact-Free Image Blending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-frame, multi-exposure camera systems in mobile devices apply a single set of calibration settings to all frames, leading to inadequate calibration for shorter exposures, resulting in color/level shifting and blending artifacts in the final output.
Innovation Solution
A method for dynamic calibration correction that involves obtaining reference and non-reference image frames, generating a mask, and applying scaling and offset coefficients to each color channel to generate a calibration-corrected non-reference image frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a single set of calibration settings is applied to all frames regardless of exposure, then the calibration process is simple and fast, but the calibration accuracy deteriorates for short exposure frames causing color/level shifting
Solution Approach 1:
The patent applies dynamic calibration by adjusting calibration coefficients based on exposure value. Instead of using a fixed single set of calibration settings for all frames, the system dynamically selects or adjusts calibration parameters according to the specific exposure conditions of each frame, thereby maintaining high calibration accuracy across varying exposure levels while managing complexity through conditional logic rather than multiple independent calibration systems
Solution Approach 2:
The patent changes calibration parameters (such as black level subtraction, white balance, lens shading correction) based on exposure value. By modifying these parameters according to the exposure conditions of each frame, the system achieves accurate calibration for both long and short exposure frames, resolving the contradiction between using a simple fixed calibration and achieving high calibration accuracy across different exposures
2Illumination intensity
If digital gains are applied to short exposure frames to compensate for underexposure, then the brightness is improved, but color/level shifting and blending artifacts are introduced
Solution Approach 1:
The patent applies calibration correction to short exposure frames before digital gain amplification. By correcting color and level shifts in advance, before the frames are amplified to compensate for underexposure, the system prevents the introduction of blending artifacts and color inaccuracies that would otherwise be magnified by the gain application
Solution Approach 2:
The patent applies preliminary anti-action by correcting color/level shifts before digital gain is applied. The calibration correction counteracts potential color inaccuracies in advance, preventing them from being amplified along with the brightness, thus maintaining color accuracy while improving illumination intensity
Data Source
AI summary
A method includes obtaining a reference image frame and a non-reference image frame and generating a mask based on the reference image frame and the non-reference image frame. The method also includes determining a scaling coefficient and an offset coefficient for each of multiple color channels of the non-reference image frame based on pixels in the reference image frame and pixels in the non-reference image frame that are identified by the mask. The method further includes applying the scaling and offset coefficients to the non-reference image frame in order to generate a calibration-corrected non-reference image frame.