Lane Detection Using Edge Expansion and Brightness Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane detection methods in vehicles are complex and prone to errors, which can lead to safety issues due to inaccurate edge detection, particularly in feature point-based methods that rely on convolution operations.
Innovation Solution
A lane detection method and system that acquires images using a vehicle-mounted photographing unit, detects positive and negative edges by comparing pixel brightness values, generates expansion images based on lane width, and combines these images for accurate lane detection, using a pixel calculation unit to determine lane width based on the camera's height and inclination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If convolution operation is performed on mask (Sobel, Robert) within ROI for edge detection, then edge detection is achieved, but calculation process becomes complex and error rate increases
Solution Approach 1:
The patent extracts only the essential edge detection functionality from complex convolution operations. Instead of applying full Sobel or Robert masks with multiple calculations, the invention extracts the core edge detection capability by comparing adjacent pixel brightness values directly, removing unnecessary computational complexity while preserving edge detection accuracy.
Solution Approach 2:
The patent inverts the traditional approach by not starting with complex convolution masks and then simplifying. Instead, it begins with the simplest possible edge detection method (direct pixel brightness comparison) and achieves effective edge detection this way, fundamentally reversing the conventional wisdom that complex operations are needed for accurate edge detection.
2Reliability
If convolution operation is performed on mask (Sobel, Robert) within ROI for edge detection, then edge detection is achieved, but error rate increases leading to safety issues
Solution Approach 1:
The patent removes the error-prone convolution operation step and extracts only the essential brightness comparison functionality. By eliminating the intermediate convolution process that introduces computational errors, the invention directly compares pixel brightness values to detect edges, thereby improving both reliability and accuracy simultaneously.
3Manufacturing precision
If pixel expansion is performed by length corresponding to lane width, then lane width coverage is improved, but calculation complexity increases
Solution Approach 1:
The patent performs preliminary calculation of the lane width in pixels before the edge expansion process. By pre-calculating the expansion distance based on known lane width specifications and camera parameters, the invention avoids complex real-time calculations during image processing, thereby maintaining high detection precision while reducing processing complexity.
Data Source
AI summary
A lane detection method includes acquiring an image in front of a vehicle using a photographing unit provided in the vehicle, detecting, from the acquired image, a positive edge (P.E.) in which a right pixel has a higher brightness value than a brightness value of a left pixel, detecting, from the acquired image, a negative edge (N.E.) in which a left pixel has a higher brightness value than a brightness value of a right pixel, generating a first expansion image by expanding a pixel by a length corresponding to a width of a lane on a right side of the positive edge (P.E.), generating a second expansion image by expanding the pixel by the length corresponding to the width of the lane on a left side of the negative edge (N.E.), and detecting the lane by combining the first expansion image and the second expansion image. In addition, a lane detection system includes a photographing unit that is provided in a vehicle to acquire an image in front of the vehicle, an edge detection unit that detects, from the image in front of the vehicle, a positive edge (P.E.) in which a right pixel has a higher brightness value than a brightness value of a left pixel and a negative edge (N.E.) in which a left pixel has a higher brightness value than a brightness value of a right pixel, an edge expansion unit that generates a first expansion image by expanding a pixel by a length corresponding to a width of a lane on a right side of the positive edge (P.E.) and generates a second expansion image by expanding the pixel by the length corresponding to the width of the lane on a left side of the negative edge (N.E.), and a lane detection unit that detects the lane by combining the first expansion image and the second expansion image.