Software detects weeds near the tracks

Researchers at the Faculty of Electrical Engineering and Computing at the University of Zagreb (Zagreb, Croatia) have developed software to detect and analyze the vegetation alongside railroad tracks.

Every year, railroads need to be maintained to ensure that the vegetation along the tracks is not dense enough to be ignited from the sparks produced by passing trains.

The most common way to eliminate such vegetation is to spray it with herbicides. Turning a sprayer on and off can be performed manually, but it is more efficient and precise to use a vision system to identify the weeds and use the data obtained from it to automatically switch the spraying nozzles on and off as required.

In doing so, the amount of herbicide, and hence the amount of environmental pollution caused by it, can be reduced. If different types of vegetation can be identified, the amount of herbicide applied can also be controlled.

In the system developed at the university, the images of the vegetation were captured by a video camera mounted on the train that was used to monitor the immediate surroundings of the railroad tracks. The detection of the vegetation was based on the color information from the images, while different types of vegetation were identified by their texture.

The researchers completed a proof of concept of the system last year, demonstrating that the software could effectively detect and classify vegetation along the tracks in real time under good lighting conditions.

The project was financed in part by the Business Innovation Center of Croatia (BICRO).

More information on other projects that the Image Processing Group at the university is involved in can be found here.

-- Dave Wilson, Senior Editor, Vision Systems Design