Variable Distortion Correction for Fisheye Panoramic Video
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wide-angle cameras used in mobile equipment capture distorted panoramic images due to fisheye distortion, which is disorienting and requires cropping to correct, leading to a trade-off between field of view and distortion, failing to provide both wide field of view and low distortion simultaneously.
Innovation Solution
A method using a hardware processor to apply variable distortion correction by determining weighting factors for each pixel coordinate in the final image frame, identifying corresponding input pixel coordinates based on a distortion model, and generating pixel values for the final image frame, allowing for different levels of distortion correction across the image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If rectilinear projection is used to correct fisheye distortion, then image distortion is reduced, but field of view must be cropped which reduces the viewing area
Solution Approach 1:
The patent applies different distortion correction strengths to different regions of the image. Central regions receive full rectilinear projection correction for straight lines, while peripheral regions use progressive correction models that maintain more of the original fisheye characteristics. This allows the image to have corrected distortion where it matters most (center) while preserving field of view at the edges.
Solution Approach 2:
The patent implements dynamic distortion correction that adapts based on the specific pixel location and distance from the image center. The correction factor varies continuously across the image frame, creating a gradient from full correction at the center to minimal correction at the periphery. This dynamic approach resolves the contradiction by making correction strength a variable parameter rather than a fixed global setting.
2Area of stationary object
If wide-angle cameras are used to capture panoramic views, then field of view is increased, but fisheye distortion increases making images disorienting
Solution Approach 1:
The patent preserves the wide field of view captured by fisheye cameras while applying localized distortion correction only where needed. The central portion of the image receives standard rectilinear correction to eliminate disorientation, while the peripheral portions maintain their distorted character to preserve the panoramic view. This selective approach keeps both the wide angle benefit and reduces distortion harm.
Solution Approach 2:
The patent effectively segments the image into multiple zones with different distortion correction characteristics. The image is divided into central, intermediate, and peripheral zones, each receiving appropriate levels of correction. This segmentation allows the system to maintain the overall panoramic field of view while correcting distortion in specific problematic regions.
3Reliability
If multiple cameras are used to provide full 360-degree coverage, then blind spots are eliminated, but system complexity and cost increase
Solution Approach 1:
The patent changes the optical parameters of a single camera system by implementing software-based variable distortion correction. Instead of adding multiple cameras to achieve complete coverage, the system modifies the image processing parameters to correct distortion in a way that maintains panoramic viewing. This parameter change approach achieves reliable coverage with a simpler single-camera system.
Data Source
AI summary
In panoramic video captured by a fisheye camera (e.g., for use in the remote operation of mobile equipment), there is a trade-off between the field of view and the level of distortion correction. In particular, to avoid the video from be disorienting, distortion correction is required. However, distortion correction applied to a very wide field of view results in an unnatural appearance of objects in edge regions, such that the image frames of the video must be cropped. A method is disclosed that varies the level of distortion correction that is applied to pixels in each image frame based on a position of those pixels in the image frame, such that less distortion correction is applied to pixels closer to edges of the image frame. This enables a very wild field of view, while preventing the unnatural appearance of objects near the edges of the image frame.