Lane-line Recognition Masking Area for 3D Object Interference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane-line recognition techniques struggle to accurately identify lane lines when a three-dimensional object, such as a preceding vehicle, is present, as they often mask too many edge points, leading to inaccurate recognition.
Innovation Solution
A lane-line recognition apparatus that includes an edge-point extractor, a three-dimensional object detector, a masking area setter, and a degree-of-belief calculator, which sets a masking area for detected three-dimensional objects and adjusts the degree of belief for edge points within this area to reduce incorrect lane-line recognition, particularly when the object meets specific conditions like being in the same lane and close to the subject vehicle.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a masking area is set to exclude edge points of three-dimensional objects, then the reliability of lane-line recognition is improved, but the quantity of usable edge points decreases
Solution Approach 1:
The patent applies local quality by differentiating the treatment of edge points based on their spatial location. Edge points within the masking area (associated with three-dimensional objects) are assigned a lower degree of belief, while edge points outside the masking area maintain their original degree of belief. This localized differentiation allows the system to preserve reliable edge points for lane-line recognition while downweighting potentially erroneous edge points from three-dimensional objects, thus resolving the contradiction between reliability and quantity of usable edge points.
2Reliability
If a large masking area is set to cover three-dimensional objects, then the reliability of excluding false edge points is improved, but the manufacturing precision of lane-line detection deteriorates
Solution Approach 1:
The patent implements dynamics by making the masking area adaptive rather than fixed. The masking area is dynamically adjusted based on the detected position and extent of three-dimensional objects. When a three-dimensional object is detected, the masking area is set to cover primarily the object itself, and when no three-dimensional object is present, the masking area is reduced or eliminated. This dynamic adjustment ensures that the masking area is large enough to exclude false edge points from three-dimensional objects while being small enough to preserve lane-line edge points, thus resolving the contradiction between false edge point exclusion and lane-line detection accuracy.
3Reliability
If edge points within the masking area are completely excluded, then the reliability of recognizing lane lines is improved, but the measurement precision of lane line position deteriorates
Solution Approach 1:
The patent applies parameter changes by modifying the degree of belief parameter for edge points within the masking area rather than completely excluding them. Instead of setting the degree of belief to zero for edge points in the masking area, the system assigns a lower but non-zero degree of belief. This parameter modification allows edge points from three-dimensional objects to be downweighted in the recognition process while still permitting edge points that genuinely belong to lane lines (even within the masking area) to contribute to the recognition result, thus resolving the contradiction between recognition reliability and position measurement precision.
Data Source
AI summary
An apparatus for recognizing a lane line. In the apparatus, when a three-dimensional object lies in the same lane as a subject vehicle and a distance between the three-dimensional object and the subject vehicle is small in an image acquired by an image capture unit, a masking area setter sets a masking area that is partially truncated at or near a lower end of the three-dimensional object in the image. A degree-of-belief calculator is configured to, for each of the edge points extracted by the edge-point extractor, calculate a degree of belief that the edge point is on the lane line. The degree of belief when the edge point is in the masking area is set less than the degree of belief when the edge point is outside the masking area. A lane-line recognizer is configured to recognize the lane line based on the degrees of belief calculated for the edge points.