Lane Detection via Spatiotemporal Image Tracing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane detection techniques fail to accurately trace driving lanes in the presence of noise similar to a lane or when a lane is effaced, leading to unreliable lane detection.
Innovation Solution
The method generates spatiotemporal images by combining line images over time and applies motion correction to linearize lane patterns, followed by Hough transformation to detect straight lines corresponding to lane patterns, and identifies lane points by tracing the coordinates and distance between these points, ensuring accurate lane detection even in challenging conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional lane detection techniques are used, then the system is simple to operate, but the detection reliability fails when noise similar to lane or effaced lane is present
Solution Approach 1:
The patent transforms 2D spatial lane detection into 3D spatiotemporal analysis by adding the time dimension. Spatiotemporal images are constructed by combining multiple frame images along the time axis, allowing the system to detect lanes not only in space but also across time, thereby improving reliability in challenging conditions while managing complexity through structured temporal processing.
Solution Approach 2:
The system performs preliminary motion correction on consecutive frame images before lane detection. By pre-aligning and stabilizing the temporal sequence of images through motion correction, the system prepares cleaner input data for subsequent lane detection, improving reliability without requiring overly complex real-time detection algorithms.
2Measurement precision
If motion correction and Hough transformation are applied, then lane pattern detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent divides the lane detection process into distinct sequential stages: motion correction of frame images, construction of spatiotemporal images, application of Hough transformation to detect straight lines, and extraction of lane information. This segmentation allows each processing stage to be optimized independently, balancing accuracy requirements with processing time constraints for each specific operation.
Solution Approach 2:
The system adjusts processing parameters dynamically, such as selecting appropriate threshold values for Hough transformation based on detected straight line characteristics, and controlling the number of frames used in spatiotemporal image construction. These parameter optimizations help reduce processing time while maintaining detection accuracy.
Data Source
AI summary
A method for detecting lanes on the road is disclosed. The method including: a phase that generates at least one spatiotemporal image by storing along the time axis at least one line image that contains pixels corresponding to at least one horizontal line established in a road video; a phase that detects, in lane patterns contained in each of the at least one spatiotemporal image, two or more lane points that correspond to the present moment; and a phase that detects lanes by tracing the two or more lane points over time in each of the at least one spatiotemporal image.