Camera Calibration Error Weighting for Multi-Camera Alignment

Solution Overview

Problem

Camera calibration accuracy is compromised when images from cameras with different optical axes have varying accuracy, leading to decreased performance in image recognition tasks.

Innovation Solution

A camera calibration information acquisition device and method that includes feature point detection, internal parameter estimation, error level calculation, and external parameter estimation using an error function that penalizes errors based on the magnitude of feature point coordinate errors, allowing for improved alignment and calibration of images from multiple cameras.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If camera calibration is performed using images from cameras with different optical axes, then the ability to handle various scenes is improved, but the accuracy of calibration decreases due to varying accuracy among cameras

Engineering Contradiction:

Improveability to handle various scenesVSAvoidcalibration accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent applies local quality by assigning different weights to feature points based on their detection accuracy for each camera. Instead of treating all feature points equally, the system calculates the magnitude of error for each feature point's coordinate detection and uses this to determine the weight. This allows the calibration to account for varying accuracy characteristics of different cameras and different feature points, thereby maintaining high calibration accuracy while handling diverse scenes captured by multiple cameras with different optical axes.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent changes the parameter of weight assignment in the calibration process. By introducing a weight calculation mechanism that depends on the magnitude of error in feature point coordinate detection, the system dynamically adjusts the importance of each feature point during calibration. This parameter change enables the system to compensate for variations in camera accuracy and maintain robust calibration across different camera types and scenes.

Inventive Principle:

Principle #35Parameter changes

2Reliability

If the error in feature point coordinate detection is large, then the robustness of calibration is improved, but the precision of external parameter calculation decreases

Engineering Contradiction:

Improvecalibration robustnessVSAvoidexternal parameter calculation precision

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent implements feedback by using the calculated magnitude of error in feature point coordinates to adjust the weight assigned to those feature points in the external parameter calculation. The error information feeds back into the calibration process, allowing the system to automatically adjust the influence of each feature point based on its detection quality. This feedback mechanism ensures that feature points with large errors have reduced weight, preventing them from degrading the overall precision of external parameter calculation while maintaining robustness.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent applies local quality by assigning different weights to feature points based on their detection accuracy for each camera. Instead of treating all feature points equally, the system calculates the magnitude of error for each feature point's coordinate detection and uses this to determine the weight. This allows the calibration to account for varying accuracy characteristics of different cameras and different feature points, thereby maintaining high calibration accuracy while handling diverse scenes captured by multiple cameras with different optical axes.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS11758110B2Camera calibration information acquisition device, image processing device, camera calibration information acquisition method, and recording medium

Publication Date: 2023.09.12 NEC CORP

AI summary

A camera calibration information acquisition device acquires captured images of a camera calibration target from two or more cameras, detects, from each image, a coordinate of a feature point in the image, calculates—an internal parameter for each camera by using the feature point, calculates, for each camera, a magnitude of error in the coordinate of the feature point, calculates a value for an external parameter of the cameras by using the magnitude of the error, the coordinate of the feature point, and an error function set so that a penalty for error in calculating the external parameter decreases as the magnitude of the error in the coordinate of the feature point in the image increases.