Camera-Based Intersecting Road Detection Using Road Width Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting intersecting roads face challenges such as high costs and incorrect detection due to GPS misalignment or lack of clear road partition lines, making it difficult to identify potential collision hazards at intersections.
Innovation Solution
A processor-based system that calculates and estimates road widths at multiple positions using image segmentation to detect intersecting roads by comparing actual and estimated widths, determining the presence of an intersecting road when the calculated width exceeds the estimated width.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If GPS and map information are used to detect the intersecting road, then the intersecting road detection coverage is improved, but the cost increases and detection accuracy deteriorates due to GPS misalignment
Solution Approach 1:
The patent replaces the GPS-based mechanical positioning system with an optical image processing system. By using the camera to capture road images and analyzing road pixel widths through image processing algorithms, the system substitutes the GPS mechanism with a visual detection mechanism that is not susceptible to GPS misalignment issues.
Solution Approach 2:
The patent creates a digital copy of the road geometry by capturing images and extracting road pixel information. Instead of relying on GPS coordinates, the system builds a visual representation of the road and its intersections by analyzing the captured image data, effectively copying the physical road structure into digital pixel space for analysis.
2Device complexity
If road partition lines are used to detect the intersecting road, then the detection method is simple, but the detection accuracy deteriorates when clear partition lines are not provided
Solution Approach 1:
The patent extracts road boundary information directly from the image data by analyzing road pixel widths, separating the detection task from reliance on pre-marked partition lines. By taking out the road geometry information directly from the visual field and analyzing it through width calculations, the system eliminates the need for artificial partition line markers.
Solution Approach 2:
The patent changes the detection parameter from the presence/absence of partition lines to the width of road pixels. By measuring how the road pixel width changes at different positions and angles, the system transforms the detection problem from line recognition to dimensional analysis, making it independent of road marking quality.
3Measurement precision
If the road width is calculated at multiple positions and compared with estimated width, then the intersecting road detection accuracy is improved, but the calculation complexity increases
Solution Approach 1:
The patent segments the road detection task by calculating widths at multiple discrete positions (first, second, and third positions) along the road. By dividing the continuous road into discrete measurement segments and analyzing width variations at each segment, the system achieves accurate intersection detection through systematic subdivision of the detection space.
Data Source
AI summary
A intersecting road detection device calculates the width of road pixels at first and second positions in a traveling direction of a host vehicle based on the road pixels corresponding to roads included in a host vehicle front image, calculates the width of the road pixels at a third position in the traveling direction of the host vehicle based on the road pixels, estimates the width of the road pixels at the third position based on the width of the road pixels at the first and second positions by assuming that an intersecting road intersecting a road on which the host vehicle is travelling is not included in the image, and determines that the intersecting road exists at the third position when calculated width of the road pixels at the third position is greater than estimated width of the road pixels at the third position.