Fiducial Marker Calibration for Fast Monocular Pose Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in accurately estimating object distance and angle using dynamic monocular cameras due to processing resource consumption and time constraints, particularly in dynamic environments, and there is a loss of region of interest (ROI) due to the distance between the camera and objects within its field-of-view (FOV).
Innovation Solution
A system and method that utilizes an image capture device to obtain a calibration image with a marker at a known distance, extracts features, determines calibration parameters, and calculates object distance and angle by capturing images at different angles, employing algorithms for feature detection and matching to estimate 3D positions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing systems calculate yaw and roll angles using complex algorithms, then measurement precision is improved, but processing time and resource consumption increase
Solution Approach 1:
The system pre-calculates and stores calibration parameters (intrinsic matrix and distortion coefficients) during a calibration phase using a calibration marker. These pre-computed parameters are then reused for subsequent angle and distance estimations, eliminating the need to perform complex calculations in real-time operation.
Solution Approach 2:
The patent uses a calibration marker (such as an AprilTag) that creates a standardized, reproducible reference pattern. By capturing images of this known marker at known positions and angles, the system creates a reference model that can be copied and applied to estimate parameters of unknown objects, replacing complex real-time calculations with pattern matching against the stored reference.
2Adaptability or versatility
If a dynamic monocular camera is used for image capture, then adaptability is improved, but measurement precision deteriorates due to distance estimation challenges
Solution Approach 1:
The calibration marker serves as an intermediary object between the camera and the target. By first measuring the marker's position and dimensions in the image, and knowing its real-world dimensions, the system calculates calibration parameters that mediate the transformation from image coordinates to real-world coordinates. This intermediary measurement enables accurate distance estimation for subsequent objects without requiring direct complex measurements.
Solution Approach 2:
The system changes the parameter state by transitioning from unknown parameters (when capturing images of unknown objects) to known parameters (when capturing the calibration marker with known dimensions and positions). This parameter transformation establishes reference values that enable accurate measurements in dynamic conditions.
3Measurement precision
If complex algorithms are used for calculating object attributes, then measurement precision is improved, but device complexity increases
Solution Approach 1:
Complex calibration computations are performed once during an initial setup phase, and the results are stored as fixed parameters. This preliminary action eliminates the need for repeated complex calculations during normal operation, reducing the processing burden while maintaining measurement precision.
Solution Approach 2:
Instead of implementing complex real-time calculation algorithms, the system copies the calibration relationships established during the calibration phase and applies them directly to measure unknown objects. This copying approach replaces complex computational algorithms with simpler parameter application and pattern matching.
Data Source
AI summary
A system and method for estimating object distance and/or angle from an image capture device may obtain an image including a fiducial marker captured by an image capture device; extract a pixel width and/or a pixel height of the fiducial marker in the image space of the image; determine, based on the pixel width and/or the pixel height, a known width and/or a known height, and one or more calibration parameters, a distance of the fiducial marker from the image capture device; estimate, based on the pixel width, the pixel height, the known width, and the known height, a yaw angle of the fiducial marker with respect to the image capture device; determine, based on a slope of a positional vector of the fiducial marker, a roll angle of the fiducial marker with respect to the image capture device; and/or provide the distance, the yaw angle, and/or the roll angle.