Lane Line Estimation Using Width Constraints and Dispersion Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane line estimating apparatuses struggle to accurately detect lane lines on express highways, particularly double white lines and tracks that can be erroneously recognized due to brightness differences and reflection light from puddles, leading to poor detection precision and unstable driving control.
Innovation Solution
A lane line estimating apparatus that includes an imaging unit, a storage unit for lane-line width data, a lane-line candidate point setting unit, a curve approximation processing unit, and a lane-line type estimating unit, which detects lane-line detection points, sets candidate points based on lane-line width, and uses curve approximation to estimate lane-line types by analyzing dispersion, thereby accurately identifying true lane lines even in complex scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If edge detection based on brightness difference is used to detect lane lines, then detection speed is improved, but measurement precision deteriorates due to erroneous recognition of double white lines and tracks as lane lines
Solution Approach 1:
The patent segments the lane line detection process into multiple stages: initial edge detection, candidate point identification, and verification through geometric relationships. By dividing the detection process, the system maintains fast initial detection while adding precision through subsequent verification steps that distinguish true lane lines from false targets like double white lines and tracks.
Solution Approach 2:
The patent implements feedback mechanisms where detected candidate points are verified against expected geometric relationships and patterns. The system uses feedback from multiple detection passes and consistency checks to confirm whether detected edges represent actual lane lines, thereby improving measurement precision while maintaining detection speed through intelligent filtering.
2Device complexity
If simple edge detection is used, then device complexity is reduced, but measurement precision deteriorates due to inability to distinguish true lane lines from pseudo lines
Solution Approach 1:
The patent applies preliminary actions by first detecting candidate edge points, then systematically verifying each candidate against geometric constraints and patterns before finalizing lane line identification. This preliminary verification process adds precision without requiring complex hardware, using computational checks to filter false positives while maintaining relatively simple device architecture.
3Ease of operation
If lane line detection is performed without considering lane-line width variations, then ease of operation is improved, but measurement precision deteriorates due to inability to identify correct lane lines among multiple candidates
Solution Approach 1:
The patent applies local quality by considering lane-line width variations at different locations and orientations. The system adjusts detection parameters and verification criteria based on local characteristics of the road environment, allowing it to maintain ease of operation through automated adaptation while improving measurement precision through location-specific optimization of detection sensitivity and geometric verification.
Data Source
AI summary
A fixed data memory stores data of a normal lane width between lane lines and a narrow lane width between inner guide lines of double white lines. A lane-line candidate setting section detects lane-line detection points on both sides of a driving lane using a captured image, and sets lane-line candidate points on the opposite lane lines at spaces of the widths and therebetween, using the detection points as starting points. A curve approximation processing section sets virtual lines on both sides of the driving lane from a curve approximation equation obtained using the detection and candidate points. A lane-line position setting section obtains dispersions of the candidate points to the left and right virtual lines, and estimates the type of at least one of the left and right virtual lines.