Camera-based visibility measurement for challenging weather conditions
Joanneum Research Digital

In air traffic management and monitoring of critical infrastructure, the exact description of the near-surface atmospheric state – and thus the visibility – is essential for situation awareness and any further weather forecast. Overcoming the current drawbacks of subjective reports from human observers, visIvis presents an innovative solution to automatically derive visibility measures from standard cameras using a vision-based approach.

The experts at Joanneum Research Digital have developed an image processing method that uses images from conventional, static-mounted digital cameras to determine the visibility of a site by combining visibility detectors and methods (e.g. feature-based methods or de-hazing methods). The operator can use existing cameras, both webcams and professional industrial cameras. Based on training data – which maps good and poor visibility conditions – the data required for visibility-measurement-suitable image areas is found automatically (such as terrain formations or buildings). At the same time, the optimal parameters for the detection of these image areas are determined. With the evaluations, a system user can very quickly record the currently prevailing visibility situation.

The latest development of visIvis also enables the monitoring of snow conditions on critical infrastructure (roads or flight paths). In addition to the primary application of visibility analytics and with a relatively low amount of training data, the visibility system can be converted into a system for snow detection using transfer learning methods, thus generating added value at the same camera location.

Booth: 4094

Back to News