Disaster monitoring

Forestry application

Maritime application

Traffic monitoring

Repository site monitoring

Season monitoring

EOFrame

Project results

Objectives

The traffic monitoring application is based on the processing and interpretation of video sequences captured from an aircraft or a high building, tower etc. The general objective is to develop:

1. An automated image alignment and geocoding method which compensates for the camera motion and enables object extraction in case of a moving platform

2. An automated procedure for the extraction of each traffic object (car, bus) by its movement from a series of consecutive images.

3. An automated tracking procedure which follows each object throughout the sequence and outputs its location and velocity as a function of time.

4. A user interface which enables interactive operation of the system and facilitates the extraction of the parameters required by the user

An effort will be made also to identify smaller objects (pedestrians, bicyclists) within the limitations of the image resolution.

Traffic objects are identified from a digital video sequence. The frame shown is captured from a helicopter at 200 meters.

Results

Two imaging flights with a helicopter took place in June 2002 over the downtown area of Helsinki. This material has been used for the method development and testing during the project. The following video sequences demonstrate the data processing chain in case of a moving camera platform:

• 200_20sec.wmv [6 MB] - A raw video sequence from Erottaja, Helsinki, distance 200 meters.

• 200mosaic.wmv [4 MB] - A video sequence stabilized by aligning all the subsequent frames with the first one using a set of stationary ground control points selected at the street surface level.

• 200MVOdetection.wmv [3 MB] - A video sequence illustrating only the moving traffic objects. In addition, the whole sequence has been geocoded (aligned) with an existing city map.

Another imaging campaign took place in summer 2005. Five intersections in Helsinki downtown area were monitored from the rooftop or the upper floor of a nearby building. No stabilization was required in this case. The following video sequences demonstrate the data obtained at short distance and with a non-vertical view:

• Kamppi455-620.wmv [2.4 MB] - A raw video sequence from Kamppi intersection, Helsinki

• Kamppi455-620MVO.wmv [2.6 MB] - A video sequence illustrating the moving traffic objects only

• Kamppi455-620Speeds.wmv [2.8 MB] - A composite video sequence illustrating the original view and the traffic objects together with their velocity vectors.

In both cases, the results will be stored in the video format, which enables various post-processing methods according to the application. The user may e.g. restrict the area of interest within the scene and concentrate on specific trajectories or monitor those vehicles stopping during the sequence.