International Society for Photogrammetry and Remote Sensing
Commission III - Photogrammetric Computer Vision and Image Analysis
Working Group III / 5 - Road Extraction and Traffic Monitoring, 2004 - 2008

Annual reports

Annual Report WG III/5 for year 2006

Major focus of WG's III/5 activities in 2006 was put on the organization, compilation and editing of an ISPRS Journal Theme Issue on "Airborne and Space-borne Traffic Monitoring" (ISPRS Journal Vol. 61, Issues 3-4). This theme issue is devoted to the particular topic "traffic monitoring" of WG III/5 and indicates the future importance of this field to Photogrammetry and Remote Sensing.

Details concerning the Theme Issue's scope, organization and papers are reflected by the editorial:
Mobility and transportation are indispensable pillars of modern society and economy. Traffic, however, is also a source of environmental pollution and a cause of injuries and deaths. Smart traffic management attempts to optimize vehicle throughput while still mitigating these adverse effects. And - traffic keeps on changing, be it on short or long time scales. Road construction, major events, or disasters require rapid reaction. Political developments, like the expansion of the European Union have significantly changed the traffic scenarios in the member states. Therefore, data and models for traffic management have to be continuously updated.
Today's road systems are equipped with a suit of sensors for monitoring traffic status: induction loops, overhead radar sensors, video systems are the most prominent examples. They all deliver accurate, reliable, timely, yet merely point-wise measurements. Airborne and spaceborne imaging systems on the other hand give us synoptic views of complex traffic situations and the associated context. These data are complementary to the one of the road sensors and can either be used in research for improving traffic models or as information source for operational monitoring systems. The context information delivered by imaging systems is important in non-standard situations, like disasters or major events.
With the recent advances in sensor technology, the automatic detection, characterization and monitoring of traffic using airborne and spaceborne data has become an emerging field of research. Approaches for vehicle detection and monitoring include not only video cameras but nearly the whole range of available sensors such as optical aerial and satellite sensors, infrared cameras, SAR systems, and airborne LIDAR. Although airborne cameras are already in use and seem to be an obvious choice, satellite systems have entered the resolution regime required for vehicle detection. Sub-metric resolution is available in the optical domain. Interferometric radars are able to spot vehicles even at night-time and under bad weather conditions. The first two satellites, the German TerraSAR-X and the Canadian Radarsat-2, to be equipped with a vehicle detection capability are ready for launch. Therefore, we believe that this special issue comes at the right time to give an overview of state-of-the-art approaches for detection, tracking, and velocity estimation of moving objects based on air- and spaceborne remote sensing systems.

After peer-reviewing the submissions in a team of 18 reviewers, ten papers have been selected for publication and grouped according to the sensor technologies used: most notably RGB cameras, infrared sensors, and Synthetic Aperture Radar (SAR) systems.
The theme issue starts with an overview paper of Toth & Grejner-Brzezinska. They present and evaluate different methods for acquiring traffic parameters from airborne optical and LIDAR data and add an interesting discussion about limitations and potentials of these methods. Then, Reinartz et al. focus on accuracy aspects that must be met during data acquisition and data processing. They analyse the complete processing chain for deriving traffic parameters from photogrammetric image sequences using semi- and fully-automatic techniques. A system for automatic tracking of vehicles in aerial video images is presented by Karimi et al. They use optical flow as basic algorithm for tracking, yet combine it with a scale-space approach to get appropriate initial values for tracking and for solving the employed differential equations. Finally, the group of approaches exploiting the visual spectral range is concluded by the article of Niu. With the incorporation of geometric deformable models, he employs a common framework to simultaneously extract roads ad well as vehicles.
The paper of Kirchhof & Stilla introduces an approach for automatic vehicle detection and tracking in aerial oblique-view infrared videos. Infrared sensors are in particular appealing for traffic applications because of their night-imaging capabilities.
The final group of papers comprises those approaches which rely on Radar and SAR data. Koch et al. present a fully Bayesian approach for tracking ground-moving targets extracted from airborne Radar data. Their approach shows that, through tracking of multiple hypotheses in multiple views, frequently used tracks ("roads") can be recovered and partially occluded areas can be bridged. The paper of Suchandt et al. describes the first spaceborne SAR experiment ever for the detection of vehicles and estimation of their velocity. SAR data acquired during the Shuttle Radar Topography Mission (SRTM) have been processed with a number of modifications compared to standard SAR processing to extract vehicles from along-track interferometric SAR data. The validity of the experiment has been confirmed by ground-truth measurements during a SRTM pass. While this paper focuses on the fundamental issues of spaceborne SAR processing, the article of Meyer et al. deals with aspects of theoretical and empirical performance characterization of so-called Ground Moving Target Indication (GMTI) approaches when applied to data as, for instance, expected from TerraSAR-X. A special feature of the GMTI approaches developed in this contribution is the incorporation of external "a-priori" knowledge to set-up conditional probability functions, which help to better distinguish between moving and stationary objects. The line of spaceborne GMTI is continued by the article of Bethke et al. It outlines the TRAMRAD-project that addresses a visionary concept showing how air- and spaceborne traffic monitoring could be realized in the future. Based on the experience of the first phases of the project, which were devoted to spaceborne traffic monitoring, the authors present first studies on a future airborne SAR system (F-SAR) with specific modes for GMTI.
Although not dealing with vehicle traffic in particular, we included also work on ship traffic detection into this theme issue. Especially with the launch of the ENVISAT satellite, monitoring ship traffic from space has gained much attention over the past years and is partly brought to practice. Due to the relatively low resolution of ENVISAT ARAR images Lopez-Martinez et al. developed a wavelet-based spot detection scheme to extract ships from polarimetric SAR data of open sea as well as coastal areas.

The articles illustrate the broad variety of approaches and methods for air- and spaceborne traffic monitoring. They also confirm the trend that modeling and analyzing dynamic processes are gaining increasing attention in high resolution remote sensing. In other words: Photogrammetry & Remote Sensing continue their way towards the full exploitation of the 4th dimension, time.

Stefan Hinz
Richard Bamler
Uwe Stilla

Last changed: 2007-02-06
by Uwe Stilla