Motor Vehicle Camera Calibration Using Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing camera calibration systems for motor vehicles are computationally intensive and often insufficiently accurate, particularly in poor weather conditions and over short driving distances with few turn maneuvers, for determining the camera's orientation angles and height.
Innovation Solution
A method that determines the camera's rotation angle, pitch angle, yaw angle, and roll angle using optical flow between successive camera images, with a regular grid to find grid point correspondences and calculate epipoles, and determines camera height using a simplified optical flow from a rectangle's movement, allowing for robust calibration without complex feature extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex feature extraction and filtering methods are used to determine vanishing points and camera orientation, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent extracts only the essential geometric information (vanishing point, horizon line) from camera images without performing complex feature extraction. By focusing on the fundamental geometric properties of the road scene and using simplified algorithms to determine these elements, the system achieves adequate measurement precision while significantly reducing computational complexity compared to methods that analyze multiple feature points and apply complex filtering.
Solution Approach 2:
Instead of using complex feature extraction to improve precision, the patent inverts the approach by using simplified geometric relationships and direct calculations. The method determines camera orientation through straightforward vanishing point calculation from road markings and horizon detection, avoiding the need for complex particle filters and RANSAC algorithms while maintaining sufficient accuracy for calibration purposes.
2Measurement precision
If multiple vanishing points are determined and filtered using particle filter and RANSAC, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent extracts only the single most relevant vanishing point from the road scene rather than determining and filtering multiple vanishing points. By directly identifying the vanishing point from the convergence of road marking lines and using simple geometric relationships, the system achieves adequate precision without the time-consuming iterative filtering processes of particle filters and RANSAC algorithms.
Solution Approach 2:
The patent skips the time-consuming iterative filtering steps by directly calculating the vanishing point from observed road features. Instead of repeatedly refining estimates through particle filters and RANSAC iterations, the method rushes through the calculation by using direct geometric relationships from the captured image, significantly reducing computation time while maintaining sufficient accuracy for the calibration application.
3Measurement precision
If complex optical flow analysis is performed to determine camera orientation, then measurement precision is improved, but productivity decreases
Solution Approach 1:
The patent extracts only the essential motion information from optical flow by focusing on the dominant flow pattern indicating vehicle movement direction. Instead of performing comprehensive optical flow analysis across the entire image, the method extracts the primary flow vector from road features and uses this simplified information to determine camera orientation, achieving adequate precision while significantly improving calibration speed.
Solution Approach 2:
The patent applies partial optical flow analysis by examining only the relevant regions of the image (road surface and markings) rather than analyzing the entire scene. This selective approach provides sufficient information for orientation determination without the computational burden of complete optical flow calculation, thereby improving productivity while maintaining acceptable measurement precision for the calibration task.
Data Source
AI summary
In a method and an apparatus for calibrating a camera system of a motor vehicle, the calibration parameters comprising the rotation angle, pitch angle, yaw angle and roll angle as well as the height of the camera above the road, the rotation angle is determined from the ascertainment of the vanishing point from a first optical flow between a first and a second successive camera image, and the height of the camera is determined from a second optical flow between a first and a second, successive camera image. To determine the first optical flow, a regular grid is placed over the first camera image, correspondences of the regular grid are searched for in the second camera image, and the first optical flow is determined from the movement of the grid over the camera images.