Multi-Camera Calibration Using a Spatially Varying Pattern
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-camera systems face challenges in calibration and synchronization when cameras have unknown positions and large viewing angle differences, leading to difficulties in 3D reconstruction and volumetric video data capture.
Innovation Solution
Utilizing a spatially varying pattern displayed by a wirelessly-connected device, captured by multiple cameras to determine initial relative poses, synchronize video sequences, and adjust camera settings for improved 3D estimation and reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Structure from Motion techniques are used to determine camera poses, then 3D reconstruction can be achieved, but a considerable initial overlap in the field of view of each camera is required which limits the flexibility of camera positioning
Solution Approach 1:
A dedicated calibration object with a known spatially varying pattern is introduced as an intermediary between the cameras and the scene. This calibration object serves as a common reference that all cameras can observe and use to determine their relative poses, eliminating the need for extensive overlap between camera fields of view. The calibration object acts as a mediator that provides the necessary geometric constraints for pose estimation without requiring direct overlap between camera views.
Solution Approach 2:
The calibration object is positioned and configured in advance with a known spatially varying pattern before the actual 3D reconstruction process. By performing the calibration step preliminarily, the system establishes the relative poses of all cameras beforehand, enabling subsequent flexible camera positioning during the actual capture process without needing to maintain overlap between views.
2Ease of manufacture
If cameras are mounted on a dedicated structure with limited movement, then calibration is simplified, but the system loses flexibility and adaptability to various capture scenarios
Solution Approach 1:
The calibration object serves as a portable intermediary that can be easily positioned in different locations and orientations. Rather than requiring cameras to be mounted on a dedicated rigid structure, the calibration object provides the necessary reference framework that can be adapted to various capture scenarios, maintaining calibration simplicity while enabling system flexibility.
Solution Approach 2:
The system transitions from a static dedicated camera rig to a dynamic configuration where cameras can be freely positioned. The calibration object, with its known spatially varying pattern, enables this dynamic positioning by providing a robust reference that works regardless of camera location or orientation, allowing the system to adapt to different capture scenarios while maintaining ease of calibration.
3Adaptability or versatility
If a spatially varying pattern is used for calibration, then camera pose determination becomes possible with unknown positions, but the pattern must be within the field of view of each camera which may limit coverage
Solution Approach 1:
The calibration process is segmented into multiple steps: first, cameras are positioned to capture the calibration object with its spatially varying pattern to determine initial poses; second, the system uses these poses to reconstruct the scene. The calibration object itself can be segmented into multiple features or markers that can be detected from different angles, allowing cameras to be positioned more widely while still maintaining the pattern within the field of view during the calibration phase.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A system and method for multi-camera capture. A spatially varying pattern is displayed on a first wirelessly-connected device and captured in video sequences acquired by a plurality of cameras. Each camera is provided in a respective further wirelessly-connected device. The video sequences are processed to identify the displayed spatially varying pattern, and the spatially varying pattern is used to determine an initial relative pose of each camera.