Vehicle Detection from Satellite Images: Intensity-, Chromaticity-, and Lane-Based Method

C Hongbo, C Lu, F Zhao, LD Shen - 2009 - trid.trb.org
C Hongbo, C Lu, F Zhao, LD Shen
2009trid.trb.org
Hurricanes or other natural disasters often result in damage to and disruption of
transportation systems that are critical to relief and recovery efforts. For this reason, it is
important for the general public and elected officials to know the operating conditions of a
transportation system as soon as possible after a disaster. The goal of this study was to
develop methods for rapidly determining the operating conditions of roadways from satellite
images by detecting the presence of debris and the blockage of roads as well as vehicle …
Hurricanes or other natural disasters often result in damage to and disruption of transportation systems that are critical to relief and recovery efforts. For this reason, it is important for the general public and elected officials to know the operating conditions of a transportation system as soon as possible after a disaster. The goal of this study was to develop methods for rapidly determining the operating conditions of roadways from satellite images by detecting the presence of debris and the blockage of roads as well as vehicle flows. This paper describes efforts for automatically delineating traffic lanes, a prerequisite for determining blockages by debris, and for detecting vehicles from satellite images. Instead of relying on pixel intensity alone, this study attempted to seek additional information to improve vehicle detection rates by combining information on traffic lane locations and the color tones of vehicles. The results show that the proposed method performs well in automatic traffic lane delineation and vehicle detection.
trid.trb.org