Non-Rectilinear Camera Distortion Correction Using 3D Scene Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing non-rectilinear lenses, such as fisheye lenses, produce distorted images that require manual parameter adjustment for de-warping, which is time-consuming and impractical for large installations.
Innovation Solution
A method and device that automatically generate a 3D model of a scene by tracking the movement of object bottom portions, like feet, to distinguish between horizontal and vertical surfaces, allowing for self-learning distortion correction without manual input.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual adjustment of de-warping parameters is performed, then distortion correction quality is improved, but installation time and labor cost increase significantly
Solution Approach 1:
The camera system automatically performs distortion correction by capturing images, identifying the horizontal surface through object bottom portions, and generating a 3D model without requiring manual parameter adjustment. This self-service mechanism eliminates the time-consuming manual configuration while maintaining correction quality.
Solution Approach 2:
The system performs preliminary actions by capturing multiple images over time to learn the scene geometry and horizontal surface boundaries before final distortion correction is applied. This preliminary learning phase enables automatic parameter determination without manual intervention during installation.
2Area of stationary object
If non-rectilinear lenses are used to capture larger scene areas, then field of view is improved, but image distortion increases
Solution Approach 1:
The system changes the projection parameters dynamically by learning the specific scene geometry and horizontal surface characteristics, then adjusting the distortion correction parameters accordingly. This allows maintaining wide-angle coverage while adapting the correction to preserve scene accuracy.
Solution Approach 2:
The patent introduces a 3D modeling dimension to solve the 2D distortion problem. By creating a three-dimensional representation of the scene with learned horizontal and vertical surfaces, the system can perform distortion correction that accounts for the spatial relationships, effectively resolving the distortion while maintaining wide coverage.
3Adaptability or versatility
If multiple cameras are deployed for comprehensive monitoring, then situational awareness is improved, but total installation and configuration time increases
Solution Approach 1:
Each camera independently performs automatic distortion correction and scene learning without requiring manual configuration. This self-service capability allows multiple cameras to be deployed simultaneously without proportionally increasing installation time, as each unit autonomously adapts to its environment.
Data Source
AI summary
A method of distortion correction in an image captured by a non-rectilinear camera includes obtaining multiple images of a scene captured by the camera over time, determining where bottom portions of objects having moved over a horizontal surface in the scene are located in the images, determining a boundary of the horizontal surface in the scene based on the determined locations of the bottom portions, generating a three-dimensional model of the scene by defining one or more vertical surfaces around the determined boundary of the horizontal surface of the scene, and correcting a distortion of at least one of the images by projecting the image onto the three-dimensional model of the scene. A corresponding device, computer program and computer program product are also provided.