3D Sensor Calibration via Iterative Coordinate Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calibrating 3D position and orientation sensors require cumbersome input of approximate values for first and second coordinate transformations, which can lead to non-convergent calculations if initial values are far from actual values, especially due to the high degree of freedom in the relationship between world and sensor coordinate systems.
Innovation Solution
An information processing apparatus and method that acquires calibration information by calculating first and second transformation information using candidate values for the second transformation, allowing iterative corrections to converge without requiring external input of approximate values, utilizing image acquisition, coordinate acquisition, and estimation units to determine the position and orientation of an image sensing device on a world coordinate system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If approximate values for first and second coordinate transformations are externally input, then calibration can be performed, but the operation becomes cumbersome and calculations may not converge if initial values are far from actual values
Solution Approach 1:
The calibration apparatus automatically generates initial values for coordinate transformations using image processing and feature point detection, eliminating the need for external input of approximate values. The system serves itself by computing initial transformation values from captured images and known world coordinates of feature points, thereby improving both ease of operation and reliability of convergence.
Solution Approach 2:
The system performs preliminary image acquisition and feature point detection before the iterative calibration process begins. By pre-computing initial transformation values from images and world coordinates, the system prepares accurate starting points for the optimization algorithm, ensuring convergence without requiring external approximate values.
2Adaptability or versatility
If the degree of freedom in the relationship between world and sensor coordinate systems is high, then the sensor can measure complex positions and orientations, but the calibration becomes more difficult and may not converge
Solution Approach 1:
The patent introduces image coordinates of feature points as an intermediary element that connects the world coordinate system and sensor coordinate system. By detecting feature points in images and using their known world coordinates, the system creates a bridge that simplifies the calibration process even when the sensor has high degrees of freedom, making the calibration tractable while preserving full measurement capability.
Solution Approach 2:
The patent replaces manual input of approximate calibration values with an automated image processing system. Instead of mechanically entering initial values, the system uses optical imaging, feature detection, and coordinate transformation algorithms to automatically generate initial values, thereby reducing calibration complexity while maintaining adaptability to complex sensor configurations.
Data Source
AI summary
A calibration information calculation unit (540) calculates a plurality of candidates of first coordinate transformation information using a plurality of candidates of second coordinate transformation information, a sensor measured value, and the position and orientation of a video camera (100) on the world coordinate system. The calibration information calculation unit (540) then calculates a piece of first coordinate transformation information by combining the plurality of calculated candidates. Then, the calibration information calculation unit (540) makes iterative calculations for correcting calibration information using a candidate of the second coordinate transformation information and the first coordinate transformation information as initial values of the calibration information.