Image Calibration via Road Area Feature Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems in mobile objects face challenges in performing appropriate calibration due to misaligned installation of imaging devices and uneven device performance, leading to inaccurate image processing for driving assistance.
Innovation Solution
An image processing device that acquires images from multiple directions, extracts feature points, detects road areas, and calibrates images based on these points to align the coordinate systems, using a combination of feature point extractors and calibrators to correct misalignments and performance variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If images are captured by multiple imaging devices mounted in different directions, then the coverage of surrounding environment is improved, but calibration accuracy deteriorates due to misaligned installation and device performance variations
Solution Approach 1:
The patent introduces road area detection as an intermediary step between feature point extraction and calibration. By detecting road areas in images from different imaging devices and extracting feature points specifically from these detected road areas, the system creates a common reference framework that mediates the calibration process. This intermediary road area detection enables accurate calibration despite misaligned installation by establishing a shared geometric reference (the road plane) that all imaging devices can be referenced against.
2Quantity of substance
If feature points are extracted from all detected areas, then the quantity of calibration points is improved, but calibration accuracy deteriorates due to inclusion of irrelevant or inaccurate points
Solution Approach 1:
The patent applies local quality by extracting feature points specifically from detected road areas rather than from all areas in the image. The feature point extraction process is localized to regions identified as road surfaces, which provides geometric constraints (the road plane) that are particularly useful for calibration. This selective extraction based on local area detection ensures that feature points are obtained from regions with known geometric properties, improving calibration accuracy while maintaining sufficient quantity of calibration points.
Data Source
AI summary
An image processing device includes: an acquirer configured to acquire a first image obtained by imaging in a first direction and a second image obtained by imaging in a second direction other than the first direction; an extractor configured to extract feature points from the first and the second image acquired; a first and a second detector configured to detect a road area included in the first and second image; a first extractor configured to extract a feature point in the road area detected by the first detector out of the feature points extracted as a first feature point; a second extractor configured to extract a feature point in the road area detected by the second detector out of the feature points extracted as a second feature point; and a calibrator configured to calibrate the first and the second image based on the first and the second feature point.