Vehicle Lane Boundary Estimation From Parallel Road Edges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately recognize own vehicle division lines in front scene images, especially when they are obscured by factors like backlight or unclearness, and fail to identify their presence at points where they do not exist, such as central intersections, hindering the development of sustainable transportation systems.
Innovation Solution
An outside recognition system that includes a front scene image acquirer, a travel path recognizer with a classifier, a parallel edge extractor, a feature ratio parameter calculator, and an estimator to identify and estimate the position of own vehicle division lines using feature ratios and stored parameters, even when the lines are not visible, by extracting parallel edges and utilizing geometrical relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If own vehicle division line recognition is performed based on image processing alone, then the system is simple, but recognition accuracy deteriorates when lines are obscured or non-existent
Solution Approach 1:
The patent combines image processing results with distance image data to recognize own vehicle division lines. When the classifier cannot detect division lines in the front scene image, the parallel edge extractor uses distance information to identify parallel edges and estimate division line positions, merging multiple data sources to maintain recognition accuracy.
Solution Approach 2:
The patent introduces distance images as an intermediary to bridge the gap when direct visual detection fails. The distance image processor provides depth information that enables the parallel edge extractor to infer division line positions even when they are not visible in the front scene image, acting as a mediator between the camera and the recognition system.
2Adaptability or versatility
If the system uses only visible division lines for travel path recognition, then the processing is simple, but the system fails at intersections where division lines do not exist
Solution Approach 1:
The patent performs preliminary extraction of parallel edges from distance images before attempting to recognize own vehicle division lines. This preliminary action prepares the system to handle cases where division lines are not visible, allowing it to adapt to intersections and other scenarios without requiring complex real-time decision-making.
Solution Approach 2:
The patent implements a dynamic recognition system that switches between different methods based on detection conditions. When division lines are visible, the system uses direct image processing; when they are not visible, it transitions to using parallel edge extraction from distance images, making the system adaptable to varying road conditions.
3Reliability
If parallel edge extraction is used to estimate division line positions, then recognition is possible without visible lines, but the calculation complexity increases
Solution Approach 1:
The patent segments the recognition process into distinct modules: classifier for direct detection, distance image processor for depth information, and parallel edge extractor for estimating division line positions. This segmentation allows each module to handle specific aspects of the problem independently, managing complexity through modular design while maintaining reliability.
Data Source
AI summary
A travel path recognizer 5 includes: a classifier 50 that classifies photographic objects in a front scene image, and generates an edge image; a parallel edge extractor 51 that extracts an own vehicle division line edge and a plurality of parallel edges from the edge image; a feature ratio parameter calculator 52 that calculates a value of a feature ratio parameter correlated to a feature ratio between a comparison distance between a reference edge and a comparison edge and an own vehicle division line distance between the reference edge and the own vehicle division line edge along a horizontal reference line; and an own vehicle division line estimator 54 that estimates, based on positions of the reference edge and the comparison edge at a current time and a value of the feature ratio parameter at a time earlier than the current time, a position of an own vehicle division line.