 |
|
|
|
|
|
|
|
|
|
|
|
|
 |
|
|||||||||||
 |
 |
 |
|
|||||||||
 |
|
|||||||||||
 |
 |
 |
|
|||||||||
 |
|
|||||||||||
 |
 |
 |
|
|||||||||
 |
|
|||||||||||
 |
 |
|
||||||||||
 |
|
|||||||||||
 |
 |
 |
|
|||||||||
 |
|
|||||||||||
 |
 |
 |
|
|||||||||
 |
|
|||||||||||
 |
 |
 |
|
|||||||||
 |
|
|||||||||||
 |
|
|||||||||||
 |
|
|||||||||||
Automated Roadway Feature Extraction from Remotely Sensed High-Resolution Satellite Imagery
Siamak Khorram and Xiaolong Dai
Funded by the National Cooperative Highway Research Program (NCHRP) Project 15-15, the National Research Council. In Cooperation with NC Supercomputing Center and Department of Civil Engineering, NCSU.
OBJECTIVE The primary objective is to analyze, evaluate, and develop automated remote sensing technologies to extract roadway inventory features from high-resolution satellite images.
INTRODUCTION According to the U.S. Department of Transportation, there are over eight million lane kilometers of different classes of roadways nationalwide. Ideally, agencies would like to have a complete, up-todate, and sufficiently accurate inventory that can be easily queried for information. However, collection and presentation of roadway inventory data under most current practices are tedious, time consuming, and costly. Satellite remote sensing technologies offers significant potential for automated roadway inventory.
APPROACH The roadway inventory features were first characterized as cover-type features, shape-type features, and measurement-type features. The technological status for extraction of these different features were then analyzed and evaluated. The technological requirements and algorithms for automated extraction of these features from remotely sensed imagery were explored. The roadway features under consideration include not only the roadway network itself but also other features, such as cover-type features (e.g., material types and land cover) and measurement-type features (e.g., road width and road curvature) used in practice. Based on the Thin and Robust edge detection algorithm, we developed a new system, called Automated Roadway Centerline Extraction System (ARCES), to extract roadway edges, centerlines, and roadway width, as shown in the flowchart.
RESULTS Roadway inventory feature extraction from high-resolution satellite imagery is expected to reduce the time and cost of roadway inventory. A variety of automated techniques for extraction of roadway features from imagery are currently available, but are usually image-dependent, feature-dependent, resolution-dependent, and SNR-(signal to noise ratio) dependent. For extraction of cover-type roadway features, the classification-based techniques are advantageous over the edge detection-based techniques. Any image georeferencing process causes positional errors. These errors can be checked using Differentiated Global Positioning Systems (GPSs). The experimental results of validation using DOQQ registered images shows that the registration accuracy is within six pixels.Experimental results using simulated 1-meter resolution imagery show that the prototype system developed for extraction roadway features is robust, fully automated, and computationally efficient.
Simultaed 1-meter Resolution Image
Edges detected after Edge Refinement
Example Highway Edges
Roadway Centerline
From Imagery to Information -- ARCES: an automated approach
Contact the Webmaster Contents Copyright © 2000 NC State University
