In-Vehicle Camera Calibration via Road Surface Feature Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for calibrating in-vehicle cameras struggle to accurately distinguish feature points on the road surface from those on surrounding structures, leading to misalignment issues, especially for side cameras with varying pitch angles and feature points of different heights, which complicates precise calibration.
Innovation Solution
A calibration apparatus and method that estimates the road surface range using feature points from non-target cameras and filters out feature points of different heights from the calibration target camera's images, ensuring only road surface points are used for calibration, employing a system with feature point detection, tracking, height calculation, road surface range estimation, and sorting units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If feature points from surrounding structures (walls, guard rails, houses) are used for calibration, then the quantity of feature points increases, but calibration accuracy deteriorates due to misalignment
Solution Approach 1:
The patent extracts and selects only feature points that exist on the road surface from the plurality of feature points detected in the camera image. This is achieved by determining the road surface range and filtering feature points based on their spatial relationship with the estimated road surface, thereby eliminating feature points from surrounding structures that would cause calibration errors.
Solution Approach 2:
The patent applies different selection criteria to different spatial regions by estimating the road surface range and determining whether each feature point lies within this range. This local quality approach ensures that only feature points with appropriate geometric properties (those on the road surface) are used for calibration, while excluding feature points from structures at different heights or positions.
2Quantity of substance
If feature points of different heights are used for calibration, then the quantity of feature points increases, but calibration reliability deteriorates due to pitch angle misalignment ambiguity
Solution Approach 1:
The patent extracts only feature points that lie on the road surface by determining the road surface range and filtering the plurality of feature points accordingly. This extraction process eliminates feature points of different heights from surrounding structures, resolving the ambiguity between pitch angle misalignment and height variations.
Solution Approach 2:
The patent changes the selection parameter from simply detecting any feature point to detecting feature points with specific spatial characteristics (those on the road surface). By estimating the road surface range and using it as a filtering criterion, the system changes the quality parameter of selected feature points from generic to specifically road-surface-based, thereby improving reliability.
3Ease of manufacture
If conventional calibration method mounting camera at vehicle is used, then calibration can be performed with simple equipment, but workload increases and cannot handle load-induced parameter changes
Solution Approach 1:
The patent enables the calibration system to automatically identify and select appropriate feature points on the road surface without requiring manual intervention or specialized measurement equipment. The system uses the camera's own images and automatically estimates the road surface range to perform calibration, making the process self-sufficient and adaptable to different vehicle load conditions.
Data Source
AI summary
There are provided a calibration apparatus and method capable of obtaining a high-precision calibration result by taking out only feature points on a road surface from images captured by an in-vehicle camera which is a calibration target. A road surface range is estimated by using feature points appearing in an image(s) captured by a camera (a front camera 102 or a rear camera 103) that is not a camera on which the calibration is executed (a side camera 106); and the calibration is executed by using only feature points existing within the estimated road surface range from among the feature points appearing in an image(s) captured by the camera on which the calibration is executed.