Road Line Recognition Using Distance-Adaptive Hough Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road line recognition systems face challenges in accurately detecting road lines, especially in complex road shapes, adverse weather conditions, and distinguishing between road lines and other high luminance features like pedestrian crossings, leading to potential misclassification and reduced detection reliability.
Innovation Solution
A road line recognition apparatus that uses imaging and image processing to detect road lines by calculating distances in real space, employing Hough conversion and stereo matching to identify straight and curved lines, while adjusting search regions based on vehicle behavior and previous detections to improve accuracy and reduce false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the detection width of the high luminance portion is set to match road line width, then road line detection accuracy is improved, but pedestrian crossings with wider lines cannot be detected
Solution Approach 1:
The patent applies dynamics by making the detection width adaptive rather than fixed. The detection width is dynamically adjusted based on the distance from the vehicle to the road marking, allowing the system to accommodate varying widths of road lines and pedestrian crossings. This resolves the contradiction by enabling accurate detection of standard road lines while also detecting wider pedestrian crossings.
Solution Approach 2:
The patent changes the parameter of detection width based on distance information. By modifying the detection width parameter according to the distance from the vehicle, the system can adapt to different types of road markings (standard road lines vs. wider pedestrian crossings) without sacrificing detection accuracy for either type.
2Adaptability or versatility
If the detection width of the high luminance portion is expanded to detect wider markings, then detection coverage is improved, but thin pedestrian crossings are mistaken as roadway
Solution Approach 1:
The system dynamically adjusts detection width based on distance, preventing the expansion that would cause false detection of thin pedestrian crossings as roadway. The adaptive width ensures that detection is expanded only when appropriate (for wider markings at appropriate distances) rather than uniformly expanding which would cause false positives.
Solution Approach 2:
The patent uses feedback from distance information to control the detection width. By incorporating distance feedback into the detection process, the system can determine whether an expanded detection width is appropriate for the current scene, preventing misclassification of thin pedestrian crossings while still detecting wider markings when needed.
3Ease of operation
If road line recognition is based on luminance information, then detection simplicity is maintained, but misclassification of high luminance features occurs
Solution Approach 1:
The patent introduces distance information as an intermediary parameter to aid in road line recognition. This additional parameter acts as a mediator that helps distinguish true road lines from other high luminance features, improving reliability without significantly complicating the detection process.
Solution Approach 2:
The patent adds a new dimension (distance) to the detection process beyond just luminance. By considering both luminance and distance information, the system can more reliably distinguish road lines from other high luminance features like pedestrian crossings, resolving the contradiction between simplicity and reliability.
4Reliability
If Hough conversion is performed on all road line candidate points, then complete road line detection is achieved, but processing time increases
Solution Approach 1:
The patent applies segmentation by dividing the detection process into two stages: first identifying road line candidate points, then performing Hough conversion. This segmentation allows the system to focus computational resources on the Hough conversion step only for relevant candidate points, reducing overall processing time while maintaining detection completeness.
Solution Approach 2:
The patent performs preliminary identification of road line candidate points before performing the computationally intensive Hough conversion. This preliminary action filters out non-relevant points, so that Hough conversion is applied only to promising candidates, thereby reducing processing time while ensuring that actual road lines are not missed.
Data Source
AI summary
A road line recognition apparatus, including: an imaging section imaging a progress path of a own vehicle including a road to output a couple of images; an image processing section calculating a distance in a real space in a set region of at least an image on one side based on the imaged couple of images; and a detection section detecting a road line; wherein the detection section includes: a road line candidate point detection and conversion processing unit detecting a pixel on a road surface as a road line candidate point based on luminance and the distance with regard to the image on one side, and performing Hough conversion of the road line candidate point; a road line straight line detection processing unit detecting one straight line proper to the road line on each of a right side and a left side of the own vehicle based on at least a position or a behavior of the own vehicle between straight lines obtained by the Hough conversion; and a road line detection processing unit detecting the road line of a shape of a straight line or a curved line by recording a road line position which is a road line candidate point indicating a road line among the road line candidate points based on the detected straight line.