Lane Line Recognition via Aerial View Curve Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane line detection methods in intelligent driving technology are prone to errors due to external interference, affecting driving safety and experience, as they do not accurately determine curve fitting in real-time.
Innovation Solution
A lane line recognition method using an electronic device with a processor that captures images, performs preprocessing, and employs a sliding window approach with convolutional neural networks to accurately identify lane lines by converting images into aerial views and fitting curves, thereby reducing interference from the environment and improving detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional image processing methods or deep learning models are used for lane line detection, then the detection process can be completed, but the detection accuracy deteriorates due to external environmental interference
Solution Approach 1:
The patent transforms the original image into an aerial view image through coordinate transformation, changing the viewing dimension from front view to top-down perspective. This dimensional change allows lane lines to appear as straight lines in the aerial view, making them easier to detect and fit with polynomial curves, thereby improving detection accuracy under environmental interference
Solution Approach 2:
The patent changes the coordinate system parameters by establishing a mapping relationship between front view coordinates and aerial view coordinates. By transforming the image coordinates and performing perspective correction, the lane line parameters in the original image are converted to a new coordinate system where polynomial fitting becomes more accurate and robust to environmental variations
2Measurement precision
If polynomial fitting is performed on lane line pixel points, then curve representation is achieved, but fitting accuracy deteriorates when lane lines are not clearly identified
Solution Approach 1:
The patent performs preliminary actions including image preprocessing (gray conversion, binarization, noise filtering) and coordinate transformation to aerial view before polynomial fitting. These preliminary steps enhance the clarity of lane lines and transform curved lane lines in front view to straight lines in aerial view, providing better initial conditions for accurate polynomial fitting
Solution Approach 2:
The patent segments the lane line detection process into distinct stages: image preprocessing, coordinate transformation to aerial view, lane line pixel point extraction, and polynomial fitting. This segmentation allows each stage to be optimized independently, with the aerial view transformation specifically addressing the curvature issue that affects fitting accuracy
Data Source
AI summary
A lane line recognition method applied to an electronic device is provided. In the method, the electronic device obtains a target image comprising lane lines, and converts the target image into an aerial view of the lane lines. Once a left lane line curve and a right lane line curve are obtained by performing a curve fitting on the lane lines according to pixel points of the aerial view, the electronic device determines whether a recognition of the lane lines is accurate according to the left lane line curve and the right lane line curve.