3D Immersive Camera Calibration for Fisheye Distortion Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing immersive content systems face challenges in accurately calibrating wide-angle lenses, particularly fisheye lenses, due to significant distortions, which affect the quality of stereoscopic images and the seamless stitching of images from multiple cameras, especially when capturing 360-degree views.
Innovation Solution
A calibration process using a 3D calibration target with a known shape and grid pattern allows for the detection and compensation of distortions introduced by cameras and lenses, generating a calibration profile that can be used to correct for these distortions during playback, enabling accurate mapping of images onto a 3D viewing environment without the need for intermediate rectilinear transformations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of moving object
If fisheye lenses are used to capture wide panoramic images, then the field of view is expanded, but significant barrel distortion is introduced
Solution Approach 1:
The patent applies preliminary action by performing distortion correction during the image capture and processing stage rather than attempting to prevent distortion at the lens level. The system captures distorted fisheye images and then applies computational algorithms to correct the barrel distortion, effectively addressing the distortion problem after it occurs but before final display or stitching.
Solution Approach 2:
The patent utilizes parameter changes by transforming the image coordinates from the distorted fisheye projection to a corrected rectilinear or equirectangular projection. This involves mathematical transformations that change the spatial parameters of the image data, mapping pixels from the distorted circular field of view to a corrected rectangular or spherical representation.
2Area of moving object
If multiple cameras are used to capture 360-degree views, then the viewing coverage is increased, but image stitching becomes more difficult due to lens variations
Solution Approach 1:
The patent applies parameter changes by establishing a standardized coordinate system and projection model that all cameras in the array must conform to. Each camera's images are transformed using the same distortion correction parameters and stitching algorithms, ensuring consistent geometric parameters across all views. This standardization simplifies the stitching process by eliminating variations in projection parameters between different cameras.
Solution Approach 2:
The patent implements universality by creating a universal distortion correction and stitching framework that works across all cameras in the array regardless of minor manufacturing variations. The same calibration parameters, coordinate transformations, and stitching algorithms are applied universally to all camera inputs, making the system robust to individual camera differences and simplifying the overall stitching process.
3Manufacturing precision
If distortion correction is applied during playback, then image quality is improved, but computational complexity increases
Solution Approach 1:
The patent applies preliminary action by performing distortion correction during the initial image capture and preprocessing stage rather than during playback. The correction parameters are calculated and applied once when the image is first acquired, and the corrected images are then stored or transmitted for playback without requiring real-time computational processing. This shifts the computational burden to the capture stage where processing time is less critical.
Data Source
AI summary
Camera and/or lens calibration information is generated as part of a calibration process in video systems including 3-dimensional (3D) immersive content systems. The calibration information can be used to correct for distortions associated with the source camera and/or lens. A calibration profile can include information sufficient to allow the system to correct for camera and/or lens distortion/variation. This can be accomplished by capturing a calibration image of a physical 3D object corresponding to the simulated 3D environment, and creating the calibration profile by processing the calibration image. The calibration profile can then be used to project the source content directly into the 3D viewing space while also accounting for distortion/variation, and without first translating into an intermediate space (e.g., a rectilinear space) to account for lens distortion.