Intersection Center Line Estimation Using Crosswalk Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computer vision-based methods for recognizing intersections are prone to recognition failures due to the sensitivity of cameras to surroundings and the complexity of recognizing a single stop line, which can lead to inaccurate identification of intersection center lines.
Innovation Solution
A method and apparatus that utilize a crosswalk recognizing unit to estimate the center line of an intersection by processing road images and providing distance information from a vehicle to the crosswalk, reducing recognition failure probabilities through a combination of image processing and GPS-based location information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a sensor-based method is used to recognize intersections, then accurate information can be provided, but installation fees are high
Solution Approach 1:
The patent replaces the hardware sensor system with a computer vision-based software system using cameras and image processing algorithms. This substitution maintains the ability to recognize intersections and provide accurate information while eliminating the need for expensive sensor installation at intersection locations.
Solution Approach 2:
The system creates a virtual representation of the intersection by capturing and processing images from cameras. Instead of using physical sensors at the intersection, the system copies visual information through imaging and processes it software-based to achieve recognition, thereby reducing installation costs.
2Ease of manufacture
If a computer vision based method is used to recognize intersections, then costs are relatively low, but the cameras are sensitive to the surroundings leading to recognition failures
Solution Approach 1:
The patent segments the intersection recognition task into multiple components: detecting stop lines, identifying crosswalks, and determining intersection centers. By dividing the problem into smaller sub-tasks with specific detection algorithms for each element, the system improves overall reliability while maintaining low costs.
Solution Approach 2:
The system uses road marking patterns (stop lines and crosswalks) as intermediary elements to infer intersection locations. Instead of directly detecting the intersection center, the system identifies these intermediate road features and uses their geometric relationships to calculate the intersection center, thereby improving recognition reliability.
3Ease of operation
If a stop line is used to recognize intersections, then drivers can be informed about approaching intersections, but if another vehicle stops on the side of the road or around the stop line, a probability of recognition failure is high
Solution Approach 1:
The patent divides the intersection identification into multiple independent detection steps: first detecting stop lines, then detecting crosswalks, and finally determining the intersection center from these features. This segmentation allows the system to use multiple reference points rather than relying solely on stop lines, reducing recognition failures when vehicles obstruct stop lines.
Solution Approach 2:
The system transitions from relying on a single-dimensional stop line detection to a two-dimensional pattern recognition approach by incorporating crosswalk detection. By adding this additional spatial dimension and using the geometric relationship between stop lines and crosswalks, the system can more reliably identify intersections even when stop lines are partially obscured.
Data Source
AI summary
Provided are a method and apparatus for estimating a center line of an intersection by recognizing a crosswalk on a road input through a camera installed in a vehicle. The apparatus includes a road information providing unit which provides information about a road being traveled based on location information of a traveling vehicle; a crosswalk recognizing unit which recognizes a crosswalk based on an input image of the intersection and the information about the road and obtains a distance from the traveling vehicle to the crosswalk; and an intersection center line estimating unit which estimates the center line of the intersection based on the information about the road and the distance from the traveling vehicle to the crosswalk. Since the center line of the intersection is estimated, the apparatus and method of estimating a center line of an intersection according to the present invention can prevent traffic accidents occurring frequently at an intersection and helps indicate direction information of ‘real vehicle navigation.’