Camera Calibration via Two-View Bundle Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing camera calibration methods for head-mounted display devices (HMDs) are computationally intensive and power-consuming, especially when recalibrating rotational parameters of cameras with overlapping fields of view.
Innovation Solution
A lightweight and efficient camera calibration process using two-view bundle adjustment optimization techniques is repeated for multiple moments in time, allowing for a statistical analysis of relative rotational parameters and reducing power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional camera calibration methods are used for HMDs, then calibration accuracy can be achieved, but computational intensity and power consumption increase significantly
Solution Approach 1:
The patent divides the camera calibration process into multiple independent two-view bundle adjustment operations, each handling a specific moment in time. This segmentation allows the system to process calibration data in smaller, manageable chunks rather than processing all calibration data simultaneously, thereby reducing computational intensity and power consumption while maintaining overall calibration accuracy through statistical analysis of results across multiple moments.
Solution Approach 2:
The patent applies partial action by performing bundle adjustment only for the rotational parameters that require recalibration rather than performing complete calibration for all camera parameters. This selective approach focuses computational resources only on the necessary calibration tasks, reducing overall power consumption while maintaining sufficient calibration accuracy for the specific application needs.
2Reliability
If camera recalibration is performed frequently to maintain accuracy, then calibration reliability improves, but power consumption increases
Solution Approach 1:
The patent implements periodic action by performing bundle adjustment at multiple discrete moments in time rather than continuously. This periodic calibration approach allows the system to maintain calibration reliability by updating parameters at appropriate intervals while avoiding the excessive power consumption that would result from continuous recalibration. The statistical analysis across multiple moments provides sufficient reliability without the energy cost of constant processing.
3Measurement precision
If bundle adjustment is repeated for multiple moments in time, then statistical analysis of rotational parameters improves, but computational complexity increases
Solution Approach 1:
The patent segments the computational task into multiple identical two-view bundle adjustment operations, each processing a specific moment in time. This segmentation strategy reduces the complexity of each individual operation while the overall process benefits from statistical analysis across multiple moments. By breaking down the large computational problem into smaller, identical sub-problems, the system achieves accurate parameter estimation without the overwhelming complexity of a single monolithic calibration process.
Data Source
AI summary
Methods for performing a camera calibration process for outward-facing cameras on devices such as head-mounted display devices are disclosed. Using cameras with overlapping fields of view, relative rotational parameters of the cameras with respect to one another may be determined using an optimization technique such as a two-view bundle adjustment algorithm. A statistical analysis of the relative rotational parameters of the cameras, determined for a plurality of moments in time, may then be made to provide updated relative rotational parameters for recalibration of the cameras. A camera calibration process, such as those disclosed, may not depend on tracking points of interest over multiple moments in time, but rather on a convergence of the relative rotational parameters determined for respective moments in time.