Road Line Detection with Correction Regions for Lane Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road line detection systems face challenges in accurately identifying road lines on roads due to misclassification of road signs as road lines, especially when vehicles are changing lanes.
Innovation Solution
A road line detection device and method that adjusts confidence scores for pixels in a correction region to prioritize road line detection, using a processor to input images from a vehicle-mounted camera to a classifier, setting a correction region based on lane changes, and correcting confidence scores to enhance road line detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a classifier is used to detect road lines by calculating confidence scores for each pixel, then road line detection can be performed automatically, but road signs may be erroneously recognized as road lines reducing detection accuracy
Solution Approach 1:
The patent applies local quality by differentiating the detection process based on spatial location. A correction region is defined around the image center where lane change maneuvers are most likely to occur. Within this correction region, confidence scores are adjusted to prioritize road line detection over other objects. This localized adjustment improves accuracy in the critical area without affecting the entire image, resolving the contradiction between automated detection and precision.
2Measurement precision
If confidence scores are corrected in a correction region to prioritize road line detection, then detection accuracy improves, but system complexity increases due to additional processing steps
Solution Approach 1:
The patent implements preliminary action by pre-defining a correction region around the image center before the actual detection process. This correction region is established based on prior knowledge that lane changes typically occur in this area. By preparing this region in advance, the system can efficiently apply confidence score adjustments only where needed, rather than processing the entire image. This reduces the computational burden and system complexity while maintaining improved accuracy.
3Measurement precision
If the correction region is set based on vehicle lane change position, then detection accuracy during lane changes improves, but the system requires additional information about vehicle state
Solution Approach 1:
The patent uses an intermediary approach by introducing a correction region as a mediating concept between the vehicle's lane change state and the image processing. Instead of directly using complex vehicle state information to adjust every pixel, the system translates this information into a spatial correction region around the image center. This intermediary representation simplifies the integration of vehicle state data with image processing, reducing the overall system complexity while maintaining improved detection accuracy during lane changes.
Data Source
AI summary
A road line detection device includes a processor configured to: calculate, for each pixel of an image acquired by a camera mounted on a vehicle, a confidence score that a road line is represented in the pixel, and a confidence score that another object is represented in the pixel; set a correction region in a range assumed to include a road line in the image during changing lanes; correct, for each pixel included in the correction region, the confidence score for a road line or the confidence score for another object in such a way that the confidence score for a road line is high relative to the confidence score for another object; and detect a road line from the image based on each pixel in which the confidence score for a road line is higher than the confidence score for another object.