Linear Feature Extraction via Reference-Line Image Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for extracting horizontal road features from aerial and image sequences are time-consuming and require complex algorithms, especially for curved roads, due to the need for accurate geo-positioning and orientation data, which is challenging for ADAS applications requiring precise road information.
Innovation Solution
The transformation of orthorectified image space to reference-line referenced image space, where curved roads are represented as straight lines, allowing for easier detection and digitization of linear road features, reducing the need for complex algorithms and increasing efficiency by enabling humans to analyze more information at once.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex algorithms are used to extract linear road features from aerial orthorectified images and image sequences, then measurement precision of geo-position can be improved, but device complexity and processing time increase significantly
Solution Approach 1:
The patent introduces a reference line as an intermediary element that connects aerial orthorectified images with ground-level image sequences. This reference line serves as a common geometric framework that enables direct feature matching without requiring complex algorithms to establish geo-positioning relationships. By using the reference line as a mediator, the system achieves accurate geo-positioning through simpler geometric transformations rather than complex image analysis algorithms.
Solution Approach 2:
The patent segments the road feature extraction process into two independent parts: aerial image processing and ground-level image processing, both referenced to a common reference line. This segmentation allows each part to be processed separately with simpler algorithms, avoiding the need for complex integrated processing of multiple image types while maintaining high measurement precision through the reference line connection.
2Measurement precision
If complex algorithms are used to detect linear features on curved roads, then measurement precision can be improved, but productivity decreases due to increased processing time
Solution Approach 1:
The patent applies dynamic resampling of image data along the reference line, adapting the sampling density to the local curvature of the road. In curved sections, more samples are taken to maintain measurement precision, while in straight sections, fewer samples are sufficient, increasing productivity. This dynamic approach allows the system to maintain high detection accuracy on curved roads without uniformly increasing processing time across all road segments.
3Measurement precision
If more image data is analyzed to capture linear road features, then measurement precision can be improved, but loss of time increases due to larger data processing requirements
Solution Approach 1:
The patent extracts only the necessary image data along the reference line from the complete aerial and ground-level image sets. By taking out only the relevant portions of image data that intersect with or are adjacent to the reference line, the system maintains measurement precision for linear feature detection while dramatically reducing the total data volume that requires processing, thus minimizing time loss.
4Manufacturing precision
If manual verification of linear road features is performed on curved roads, then manufacturing precision can be improved, but ease of operation deteriorates due to difficulty in visual analysis
Solution Approach 1:
The patent dynamically adjusts the display of resampled image data along the reference line, presenting it in a linearized format that straightens curved road features. This dynamic presentation transforms the visual analysis task from examining curved features in their original geometric context to analyzing linearized features along a straight reference line, significantly improving ease of operation while maintaining verification accuracy.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method of producing linear features along a reference-line across a surface for use in a map database is disclosed. The method comprises: - generating from reference-line data representative of coordinates of said. reference- line in a geographic coordinate reference system and source images of the surface adjacent to said reference-line and associated position and orientation data in said geographic coordinate reference system, a reference-line referenced data set, wherein the reference-line referenced data set comprises a plurality of sets of image data and associated data defining a reference-line' across a surface in the geographic coordinate reference system, the sets of image data having pixels wherein a set of image data corresponds to an orthorectif ied view representation of a line section of the surface in the geographic coordinate reference system, each set of image data comprises a reference. pixel being associated with a position on the reference-line, wherein each pixel represents a surface having a position at a predefined distance from the position of the reference pixel along the line section, and wherein the line section perpendicularly crosses the reference-line at the position associated with the reference pixel; and, - post processing the reference-line referenced data set to produce linear features along the reference-line and associated locations in the geographic coordinate reference system for use in a map database.