On-board Video Camera Orientation Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calibrating the orientation of video cameras on board moving vehicles are either complex, expensive, or unable to dynamically correct orientation drifts during missions, making them impractical for mobile mapping applications.
Innovation Solution
A method using three accelerometers to measure roll and pitch angles, combined with image processing to determine camera orientation, allowing for dynamic calibration and correction of camera orientation by averaging angle estimates from multiple image pairs and using satellite/terrestrial positioning systems to calculate vehicle heading.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If precision assembly with angular markers is used to align the camera optical axis, then initial calibration accuracy is improved, but the calibration becomes unstable under vehicle vibrations and requires frequent repetition
Solution Approach 1:
The patent replaces mechanical precision assembly and angular markers with an optical field-based calibration method. Instead of relying on physical alignment components that are sensitive to vibrations, the system uses image processing of natural or artificial features in the scene to continuously determine camera orientation parameters, making the calibration immune to mechanical instability caused by vehicle vibrations.
Solution Approach 2:
The patent implements continuous feedback-based calibration by repeatedly capturing images, processing them to extract orientation information, and updating the camera orientation parameters in real-time. This closed-loop approach allows the system to compensate for any drift or changes in camera orientation during the mission, maintaining accurate calibration throughout operation rather than relying on a single initial mechanical alignment.
2Ease of operation
If a turret or gimbal is used to make camera aiming easier, then ease of operation is improved, but the device becomes expensive and still requires frequent calibration adjustments
Solution Approach 1:
The patent eliminates complex mechanical aiming systems like turrets and gimbals by using computational methods. The camera orientation is determined through image processing algorithms that calculate the position and orientation of the camera based on captured images and known positions of reference features, replacing mechanical adjustment mechanisms with software-based solutions.
Solution Approach 2:
The calibration system is self-calibrating by automatically processing images to determine camera orientation parameters without requiring external intervention or mechanical adjustment. The system uses the captured images themselves and the known positions of features in the scene to compute and update its own orientation parameters autonomously.
3Measurement precision
If a three-dimensional target on a building is used for calibration, then calibration reference is provided, but the method is restrictive and cannot correct orientation drifts during the mission
Solution Approach 1:
The patent transitions from static calibration using fixed targets to dynamic calibration using continuously captured images. The system can process images taken at any moment during the mission to update camera orientation parameters, allowing calibration to adapt to changing conditions and correct drifts in real-time without requiring the vehicle to return to specific calibration points.
Solution Approach 2:
The calibration method becomes universal by not requiring specific calibration targets or locations. Instead, it can use any scene with detectable features whose positions can be determined, allowing calibration to be performed anywhere during the mission using the camera's own field of view, making the system adaptable to various operating conditions and environments.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach provides a simple, cost-effective, and dynamic method for calibrating camera orientation, enabling continuous correction of orientation drifts and improving the accuracy of mobile mapping systems.
Implementation Method 1
The roll angle and the pitch angle of the camera are then measured by means of said accelerometers
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The method involves determining vector components for providing a movement direction of an on-board video camera between two images, expressed in a location related to the camera, from respective coordinates of matched points in two successive images. The obtained vector is projected in a horizontal plane, and an estimation of an angle between an optical axis of the camera and an axis of a vehicle, seen in the horizontal plane is deduced from the projected vector components.