TRB Pubsindex
Text Size:
x

« Prev Next »

Return to Results | New Search

Bookmark and Share

Title:

SIFT-Based Road Sign Inventory System

Accession Number:

01043529

Record Type:

Component

Availability:

Transportation Research Board Business Office

500 Fifth Street, NW
Washington, DC 20001 United States

Abstract:

Mobile mapping has become a mature methodology for road sign inventory. However, image processing and feature extraction for signs are still a study area for automation and accuracy. Current available technologies in the North America market are still largely dependent on manual assisted processing in the road network data analysis for signs. This paper aims at finding a better solution to the image processing and feature extraction problem in the road sign inventory system. The proposed system has the following properties which give better performance than the available methodologies. First, based on the experiment and statistic analysis of the road sign appearance during the field data collection, HSL color space is used to recognize signs. Raw roadway images are subsequently classified into different color bands at the beginning of the processing. This process deals with the classification of large number of road signs with different colors and shapes. Second, other than detection with conventional shape and color properties of the signs, SIFT technique is used to extract the invariant features of the detected candidate region. Recognition is accomplished by comparing those features of the candidate region and the features of the standard sign in the established database of standard signs. The system has the capability of automated detection of road signs, and performing recognition, identification, positioning analysis, and inventory management for signs. The system can be integrated into the Digital Highway Data Vehicle (DHDV) to capture the Right-of-Way (ROW) images on the roadway and analyze the image data in real-time. Preliminary experimental results are given in the paper to demonstrate the effectiveness of the proposed system.

Monograph Accession #:

01042056

Report/Paper Numbers:

07-0530

Language:

English

Corporate Authors:

Transportation Research Board

500 Fifth Street, NW
Washington, DC 20001 United States

Authors:

Wang, Kelvin C P
Hou, Zhiqiong
Gong, Weiguo

Pagination:

21p

Publication Date:

2007

Conference:

Transportation Research Board 86th Annual Meeting

Location: Washington DC, United States
Date: 2007-1-21 to 2007-1-25
Sponsors: Transportation Research Board

Media Type:

CD-ROM

Features:

Figures; Photos; References (21)

Uncontrolled Terms:

Right of way

Subject Areas:

Freight Transportation; Highways; Operations and Traffic Management; Planning and Forecasting; I72: Traffic and Transport Planning

Source Data:

Transportation Research Board Annual Meeting 2007 Paper #07-0530

Files:

TRIS, TRB

Created Date:

Feb 8 2007 5:07PM