Stereo Camera Optical Axis Alignment Using Checkerboard Vanishing Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calibrating stereo camera systems in autonomous vehicles lack efficiency and accuracy in aligning optical axes without direct mechanical measurements, particularly in real-world environments.
Innovation Solution
A method and system that calibrate intrinsic parameters of cameras by processing images from a camera and LiDAR data, extract corner points from patterns, and compute a vanishing point to adjust optical axes in parallel, using a non-transitory computer readable storage medium and a system with processing units to execute these steps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If direct mechanical measurements are used for camera calibration, then measurement accuracy may be improved, but the complexity of the calibration process and equipment requirements increase
Solution Approach 1:
The patent replaces direct mechanical measurements with an optical calibration method that uses a checkerboard pattern and vanishing point computation. Instead of using mechanical measurement tools to directly measure camera geometry, the system uses image processing of a known pattern to indirectly determine intrinsic and extrinsic parameters, thereby reducing equipment complexity while maintaining measurement accuracy.
Solution Approach 2:
The patent introduces a checkerboard pattern as an intermediary object for calibration. This pattern serves as a mediator between the camera system and the calibration process, allowing the system to compute camera parameters through image analysis of the pattern's vanishing points rather than through direct mechanical measurements, thus simplifying the overall calibration process.
2Measurement precision
If iterative adjustments are made for optical axes alignment, then alignment accuracy may be improved, but the calibration time and processing steps increase
Solution Approach 1:
The patent performs preliminary computation of vanishing points from the checkerboard pattern images before proceeding to optical axes alignment. By pre-computing these geometric features and using them to directly determine alignment parameters, the system eliminates the need for iterative adjustments, thereby reducing calibration time while maintaining alignment accuracy.
Solution Approach 2:
The patent uses the computed vanishing points as feedback to directly determine the required optical axes alignment. The vanishing point positions provide immediate geometric information about the camera orientations, allowing the system to calculate the precise alignment adjustments needed without trial-and-error iterative processes, thus reducing calibration time.
3Reliability
If multiple patterns at different orientations are disposed in the scene, then camera calibration completeness is improved, but the setup complexity and scene requirements increase
Solution Approach 1:
The patent makes the single checkerboard pattern multi-functional by utilizing it in multiple orientations and positions within the scene. The same pattern object serves multiple calibration purposes by capturing images from different angles, eliminating the need for multiple separate calibration patterns or complex multi-object setups, thus reducing setup complexity while maintaining calibration completeness.
Solution Approach 2:
The patent employs a dynamic calibration approach where the checkerboard pattern is disposed at different orientations in the scene, and the system adaptively processes images from these various configurations. This dynamic use of a single pattern in multiple states replaces the static requirement for multiple fixed patterns, simplifying the calibration setup while ensuring comprehensive camera parameter determination.
Data Source
AI summary
A method of aligning optical axes of cameras for a non-transitory computer readable storage medium storing one or more programs is disclosed. The one or more programs comprise instructions, which when executed by a computing device, cause the computing device to perform by one or more autonomous vehicle driving modules execution of processing of images using the following steps comprising: calibrating intrinsic parameters of a set of cameras; extracting corner points associated with a pattern; and computing a vanishing point based on information on the extracted corner points.