Roadside Detection Using Image Area Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lane detection systems fail to reliably identify the roadside when there are no lane markings, particularly on smaller roads that merge into unpaved areas, due to the complexity of image patterns at the transition between the road and outside spaces.
Innovation Solution
A method that divides captured images into areas, determines image features based on pixel intensity, uses prior information such as vehicle speed and yaw angle to estimate the probability of each area belonging to the road, and distinguishes 'road' and 'non-road' areas by comparing features with characteristic values, employing statistical analysis and a classifier to accurately differentiate between road and outside areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current lane detection methods are used, then lane markings can be reliably detected, but detection fails on unmarked roads
Solution Approach 1:
The patent changes the detection parameters from relying on lane marking characteristics (color, contrast, shape) to using road surface characteristics (texture, intensity distribution, geometric patterns). This parameter transformation enables the system to detect roads without markings by analyzing the statistical properties of pixel intensities and their spatial distributions, thereby achieving both reliability and adaptability across different road types.
Solution Approach 2:
The patent replaces the mechanical/visual approach of detecting lane markings with a statistical/image processing approach. Instead of searching for specific marking patterns, the system uses intensity value analysis, histogram calculations, and statistical comparisons to identify road boundaries. This substitution of detection methodology enables reliable detection on unmarked roads where traditional visual methods fail.
2Reliability
If image-based detection is used on unmarked roads, then roadside can be identified, but detection accuracy decreases due to complex image patterns
Solution Approach 1:
The patent divides the captured image into multiple regions or zones, analyzing intensity distributions and statistical characteristics separately for each region. This segmentation allows the system to handle complex image patterns by processing smaller, more manageable sections independently, then combining results to achieve precise overall roadside detection despite the complexity of unmarked road environments.
Solution Approach 2:
The patent performs preliminary analysis of image characteristics, intensity distributions, and statistical parameters before making roadside detection decisions. By pre-calculating histograms, intensity statistics, and comparing them against reference models or thresholds, the system prepares detection criteria in advance, improving accuracy when actual roadside identification is needed in complex unmarked road scenarios.
Data Source
AI summary
In a method for identifying a roadside using an image capture device mounted on a motor vehicle, at least a portion of a captured image is divided into several image areas. For each image area, at least one image feature is determined based on the intensity values of the pixels assigned to that image area. For each image area, a probability is determined, based on prior information, that the image area belongs to a lane on which the motor vehicle is currently located. Depending on the image features and probabilities of all image areas, a characteristic value is determined for each image feature of an image area that is characteristic of the lane.Each image area is assigned by comparing the image feature determined for that area with the corresponding characteristic value of a road located within the traffic area or an external area of the traffic area not belonging to the road. This assignment of image areas identifies a road edge.