Lane Marking Detection via Image Segmentation and Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated lane marking systems require extensive training data and significant computing resources due to the variability of road markings and the presence of extraneous data in images, leading to inefficiencies and inaccuracies in identifying lane markings for autonomous vehicles.
Innovation Solution
The system preprocesses images by dividing them into sub-images of straight road sections, normalizing them to align with a horizontal axis, and cropping out non-road pixels, reducing the amount of training data needed for machine learning models while emphasizing road pixels, thus improving accuracy and reducing computational requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If automated lane marking systems process complete images with extraneous data, then comprehensive road information is captured, but computing resources and processing time increase significantly
Solution Approach 1:
The patent divides the image processing task into distinct segments: first identifying the road portion in the image, then separately processing only that road region for lane marking detection. This segmentation eliminates extraneous non-road data while maintaining comprehensive road information, thereby improving processing efficiency without sacrificing detection accuracy.
Solution Approach 2:
The patent extracts and isolates the road portion from the complete image by identifying road pixels and separating them from background pixels. This extraction process removes extraneous data (buildings, trees, sky) while preserving all relevant road information, reducing computational load while maintaining lane marking detection accuracy.
2Measurement precision
If large amounts of training data are used to account for road variability, then lane marking detection accuracy improves, but training time and computational resources increase
Solution Approach 1:
The patent changes the parameter of image representation by transforming complete images into road-portions-only images. This parameter change allows the machine learning model to train on more efficient data structures that capture road variability without requiring proportionally larger datasets, reducing training time while maintaining detection accuracy.
Solution Approach 2:
The patent performs preliminary action by pre-processing images to extract road portions before feeding them to the machine learning model. This preliminary extraction of relevant information reduces the data volume requiring processing during training, thereby reducing training time while preserving all necessary road variability information for accurate lane marking detection.
3Quantity of substance
If complete images including background are processed, then all visual information is available, but the proportion of relevant road data decreases
Solution Approach 1:
The patent extracts the road portion from complete images by identifying and isolating road pixels from background pixels. This extraction increases the concentration of relevant road information in the processed data, improving the signal-to-noise ratio while maintaining the total quantity of road-related information needed for accurate lane marking detection.
Data Source
AI summary
Systems and processes can automatically identify lane markings within images through the use of a machine learning model. The machine learning model may use a reduced set of data and output an improved estimate of lane markings by applying normalized data or images to the machine learning model. Each image applied to the model can be normalized by, for example, rotating each of the images such that the depicted roads are horizontal or otherwise share the same angle. By aligning disparate images of roads, it is possible to reduce the amount of data applied to the model or to model generation, and to increase the accuracy of the machine learning model. Further, the use of normalized images by the machine learning model enables a reduction in computing resources used to apply data to the machine learning model to, for example, identify lane markings within images.