Depth Camera Calibration Using Sparse Dot Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional depth camera calibration methods using external targets like checkerboards are limited by their 2D nature, reflective surfaces, lighting conditions, and the need for complex calibration stations, leading to inaccuracies and challenges in accurately calibrating 3D depth sensors.
Innovation Solution
A method for calibrating depth cameras without external targets, utilizing an illumination source to project a pattern of dots, identifying dot locations in raw depth images, and applying an objective function to generate distortion correction parameters, enabling accurate calibration in diverse environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional external targets like checkerboards are used for calibration, then calibration can be performed with existing methods, but the calibration accuracy is limited due to 2D nature and reflective surfaces
Solution Approach 1:
The patent extracts the calibration target from the external environment and integrates it into the depth camera itself by using the camera's illumination source to project the dot pattern. This eliminates the need for separate external calibration targets and complex calibration stations, while enabling accurate 3D calibration through the camera's own optical path.
Solution Approach 2:
The illumination source serves multiple functions: it provides structured light for depth measurement and simultaneously serves as the calibration target projector. The dot pattern projected by the illumination source acts as both the measurement signal and the calibration reference, eliminating the need for separate calibration equipment.
2Reliability
If external calibration targets are used, then calibration can be performed, but lighting conditions and reflective surfaces affect calibration reliability
Solution Approach 1:
The depth camera uses its own illumination source to project the calibration dot pattern, making the system self-sufficient for calibration. The projected dots serve as the calibration target, eliminating dependence on external targets that are affected by environmental lighting and surface reflectivity. The system calibrates itself using its own active illumination.
Solution Approach 2:
The patent changes the calibration target from passive external objects (checkerboards) to active projected light patterns. By using modulated structured light with specific temporal characteristics, the system can distinguish the calibration dots from environmental reflections and achieve reliable calibration across diverse lighting conditions.
3Measurement precision
If 2D checkerboard targets are used for calibration, then calibration process is simpler, but calibration accuracy for 3D depth sensors is insufficient
Solution Approach 1:
The patent transitions from 2D checkerboard targets to 3D spatial dot patterns projected by the illumination source. The calibration target exists in three-dimensional space with known spatial relationships, enabling accurate calibration of the depth camera's 3D measurement capabilities while maintaining operational simplicity through automated pattern projection and detection.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables precise calibration of depth cameras in various lighting conditions without external targets, maintaining accuracy over the camera's lifespan and reducing the need for costly calibration stations.
Implementation Method 1
A raw depth image of illumination reflected from the environment is acquired via an optical sensor of the depth camera
Data Source
AI summary
Examples are disclosed relating to a method for calibrating a depth camera without requiring an external target. In one example, an environment is illuminated using an illumination source of the depth camera. The illumination source is configured to output modulated structured light comprising a pattern of dots. A raw depth image of illumination reflected from the environment is acquired via an optical sensor of the depth camera. Observed locations of dots in the pattern of dots are identified in the raw depth image. An objective function is applied to the observed locations of the dots in the pattern of dots in the raw image to generate a set of distortion correction parameters. A distortion corrected depth image generated based at least on translating pixel locations of pixels of the raw depth image according to the set of distortion correction parameters is output.