FNR Marker Pattern for Multi-Camera Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing calibration methods, such as chessboard calibration charts and coded targets, are inadequate for multi-camera setups in large environments due to issues like perspective distortion, low image resolution, out-of-focus targets, variable lighting, and cluttered backgrounds, which hinder accurate sub-pixel position detection and camera alignment.
Innovation Solution
The method involves generating and decoding Fourier Noise Ring (FNR) marker patterns with unique identifiers, which are robust to occlusion, shadow, and focus variations, using a two-dimensional pattern in the frequency domain with modified peripheral properties for high accuracy camera calibration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If multiple chessboard calibration charts are used in large arena, then coverage in each camera image plane is improved, but confusion between identical charts under perspective distortion worsens
Solution Approach 1:
The calibration target is segmented into multiple functional components: a coded target region with unique identifier, a grid pattern region for geometric calibration, and optional text region. This segmentation allows each region to serve its specific purpose without interfering with others, solving the confusion problem while maintaining coverage.
Solution Approach 2:
Different regions of the calibration target have different local qualities - the coded target region contains unique identifying information for differentiation, while the grid pattern region provides geometric reference points. This local differentiation allows multiple targets to be deployed without confusion while maintaining calibration accuracy.
2Loss of information
If coded targets with bar codes are used, then unique identification is improved, but performance under low resolution and out-of-focus conditions worsens
Solution Approach 1:
The calibration target utilizes multiple dimensions of information encoding: spatial frequency domain patterns, geometric grid structures, and coded identifier regions. This multi-dimensional approach ensures that identification and position detection can proceed even when one dimension is degraded by low resolution or focus issues.
Solution Approach 2:
The calibration target is a composite structure combining different pattern types (coded targets, grid patterns, frequency domain patterns) that work together. The composite nature provides redundancy - if one pattern type is difficult to detect, another can compensate, maintaining measurement precision under varying conditions.
3Ease of manufacture
If traditional calibration patterns are used, then simplicity of implementation is improved, but robustness under variable lighting and shadow conditions worsens
Solution Approach 1:
The calibration target incorporates frequency domain patterns with specific spectral characteristics that are less sensitive to lighting variations. The use of Fourier noise rings and coded patterns with defined frequency content allows the target to maintain detectability under variable lighting and shadow conditions while keeping implementation relatively simple.
Data Source
AI summary
A method of generating an image, the method including receiving a two-dimensional pattern in the frequency domain, modifying one or more peripheral properties of the two-dimensional pattern and generating the image based on the modified two-dimensional pattern.