Camera Extrinsic Parameter Verification via Perturbation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for verifying calibrated extrinsic parameter values of cameras in autonomous driving systems are limited in accuracy and applicability, failing to comprehensively check whether these values have reached optimal levels, especially considering iterative interference.

Innovation Solution

A method involving the application of multiple perturbations to calibrated extrinsic parameter values, calculation of reprojection errors, and iterative verification to determine if the values correspond to a minimum error, ensuring high accuracy and applicability in assessing optimal parameter values.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a single method is used to check calibrated parameters, then the verification process is simple, but the verification results are not comprehensive and accurate

Engineering Contradiction:
Improveverification accuracyVSAvoidverification method complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The verification process is divided into multiple independent verification methods, each targeting specific aspects of extrinsic parameter accuracy. Multiple perturbations are applied separately and their effects are evaluated independently, allowing comprehensive verification without overwhelming complexity in a single monolithic process

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The verification system integrates multiple verification methods that can handle different scenarios and parameter types universally. The same basic framework accommodates various perturbation types and verification criteria, making the system broadly applicable to different camera configurations and calibration scenarios

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

2Measurement precision

If iterative interference verification is not considered, then the verification process is faster, but it cannot check whether calibrated extrinsic parameter values have reached optimal values

Engineering Contradiction:
Improveparameter optimality verificationVSAvoidverification time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The verification process performs preliminary checks using coarser perturbations first to quickly identify obvious deviations from optimality. Only when necessary does it proceed to finer perturbations and multiple iterations, ensuring optimal verification without always requiring full iterative processing time

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The verification method dynamically adjusts the perturbation step sizes based on iteration number and verification progress. Larger perturbations are used in early iterations for quick assessment, while smaller perturbations are applied in later iterations for precise optimality verification, balancing speed and accuracy adaptively

Inventive Principle:
Principle #15Dynamics

3Measurement precision

If multiple perturbations with decreasing step sizes are applied in iterations, then the verification accuracy is improved, but the computational complexity increases

Engineering Contradiction:
Improveverification precisionVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The verification process applies perturbations selectively rather than exhaustively testing all possible parameter variations. By using a limited set of perturbation directions and stopping when convergence criteria are met, the system achieves sufficient verification precision without the computational burden of complete exhaustive search

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The perturbation step size parameter is dynamically changed across iterations, starting with larger values for rapid initial assessment and progressively reducing to smaller values for refined verification. This parameter adaptation allows the system to achieve high precision while managing computational complexity through intelligent resource allocation

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4435722A1Method for verifying calibrated extrinsic parameter values of camera, device and medium
Publication Date: 2024.09.25 NIO TECH ANHUI CO LTD
  • EP4435722A1 patent drawingFigure 1~2
  • EP4435722A1 patent drawingFigure 3~4
  • EP4435722A1 patent drawingFigure 5

AI summary

The disclosure relates to the field of computer vision technologies, and specifically provides a method for verifying calibrated extrinsic parameter values of a camera, a device, and a medium, aiming to solve the problem of being unable to check whether the calibrated extrinsic parameter values of the camera are optimal values. To this end, the method of the disclosure includes: applying a plurality of perturbations to calibrated extrinsic parameter values of a camera, obtaining at least one perturbed extrinsic parameter value based on any one of the perturbations applied, calculating first reprojection errors respectively corresponding to the calibrated extrinsic parameter values and the perturbed extrinsic parameter value, and verifying, based on a comparison between an extrinsic parameter value corresponding to a minimum first reprojection error in the first reprojection errors and the calibrated extrinsic parameter values, whether the calibrated extrinsic parameter values are optimal values. Through the above implementation, the convergence of iterative interference may be checked, and whether the calibrated extrinsic parameter values of the camera are optimal values may be verified, which has the characteristics of high accuracy and high applicability.