Stereo Camera Calibration via Parallax Offset Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing calibration methods for stereo camera distance measurement systems in vehicles rely on specific shapes or features, such as traffic signals or flat road surfaces, which limits their applicability and accuracy, especially when these features are not present or change, and they require external sensors for position measurement.

Innovation Solution

A calibration apparatus and method that calculates parallax offset using images captured during travel, without relying on specific shapes or external sensors, by analyzing the parallax and its changing amount across stationary and moving objects, allowing for stable calibration of the stereo camera system.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If calibration uses specific shapes or features (traffic signals, road surfaces), then calibration can be performed with available methods, but applicability is limited when these features are not present or change

Engineering Contradiction:
Improveapplicability of calibration methodVSAvoidstability of calibration
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent enables the calibration system to perform distance measurement calibration using multiple different features including traffic signals, road surfaces, and building facades. The system automatically detects and adapts to different feature types in the environment, making the calibration method universally applicable across various scenes without requiring specific pre-configured calibration targets.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent changes the calibration approach from relying on fixed geometric shapes to using parallax offset detection based on distance measurement variations. By monitoring how parallax values change with distance for different feature types, the system extracts calibration parameters that are invariant to the specific feature shape, thereby improving both adaptability and reliability.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If calibration relies on external sensors for position measurement, then position accuracy can be improved, but device complexity increases

Engineering Contradiction:
Improveposition measurement accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent makes the stereo camera system self-calibrating by using its own distance measurement capability to detect parallax offsets. The system measures distances to features in the environment, calculates expected parallax values based on stereo geometry, and automatically computes calibration corrections without requiring external position sensors or additional calibration equipment.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent introduces parallax offset as an intermediary parameter that connects the stereo camera's distance measurements to the calibration process. By detecting deviations in parallax values from expected geometric relationships, the system indirectly measures misalignment without directly measuring camera positions, thus avoiding the need for external position sensors.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Manufacturing precision

If precise adjustment is performed at manufacturing, then initial calibration accuracy is achieved, but long-term accuracy degrades due to temperature change and vibration

Engineering Contradiction:
Improveinitial camera alignment accuracyVSAvoidcamera arrangement stability
Core Design Contradiction:
Manufacturing precisionVSStability of the object's composition

Solution Approach 1:

The patent implements a feedback-based calibration system that continuously or periodically detects parallax offsets during vehicle operation and automatically updates calibration parameters. By monitoring distance measurements and parallax values in real-world conditions, the system compensates for drift caused by temperature changes and vibrations, maintaining long-term accuracy without requiring re-adjustment of the physical camera mounting.

Inventive Principle:
Principle #23Feedback

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 accurate and stable calibration of the stereo camera system by utilizing various images captured during travel, improving distance measurement accuracy without the need for specific features or external sensors, enhancing the system's applicability and reliability.

Implementation Method 1

two imaging devices 20, 21 for imaging a subject

Methodology Applied
Scientific EffectLight reflection: Reflection

Implementation Method 2

The distance p indicates a deviation amount of the points on the images captured by these two cameras and is called parallax

Methodology Applied
Scientific EffectParallax: Parallax

Data Source

PatentEP2616768B1A calibration apparatus, a distance measurement system, a calibration method and a calibration program
Publication Date: 2020.03.25 RICOH CO LTD
  • EP2616768B1 patent drawingFigure 1
  • EP2616768B1 patent drawingFigure 2
  • EP2616768B1 patent drawingFigure 3

AI summary

A disclosed calibration apparatus is configured to receive inputs of two reference images and plural items of parallax data, these two reference images being captured by one of the imaging devices at two locations, and the parallax data being calculated using these two reference images and two corresponding images based on positions of plural feature points common to the reference image and the corresponding image for each pair thereof on a location basis, two corresponding images being captured by another of the imaging devices at these two locations; search for plural feature points common to these two reference images; calculate parallaxes and parallax changing amounts, based on the parallax data related to the respective feature points in these two reference images, for each of the searched feature points; and calculate a correction value for the parameter based on the calculated parallaxes and parallax changing amounts.