Big data analysis in uav surveillance for wildfire prevention and management

Abstract

While wildfires continue to ravage our world, big data analysis aspires to provide solutions to complex problems such as the prevention and management of natural disasters. In this study, we illustrate a state-of-the-art approach towards an enhancement of UAV (Unmanned Aerial Vehicle) surveillance for wildfire prevention and management through big data analysis. Its novelty lies in the instant delivery of images taken from UAVs and the (near) real-time big-data oriented image analysis. Instead of relying on stand-alone computers and time-consuming post-processing of the images, a big data cluster is used and a MapReduce algorithm is applied to identify images from wildfire burning areas. Experiments identified a significant gain regarding the time needed to analyze the data, while the execution time of the image analysis is not affected by the size of the pictures gathered by the UAVs. The integration of UAVs, Big Data components and image analysis provides the means for wildfire prevention and management authorities to follow the proposed methodology to organize their wildfire management plan in a reliable and timely manner. The proposed methodology highlights the role of Geospatial Big Data and is expected to contribute towards a more state-of-the-art knowledge transfer between wildfire confrontation operation centers and firefighting units in the field.