Feature Extraction for Drone Classification

In this research an approach for extracting features from image dataset is proposed for classification vertical take-off and landing unmanned aerial vehicles. These vehicles are highly movable targets in various backgrounds while they occupy a small portion of the camera field of view. For this propose, histogram of oriented gradient of these flying objects is used on real image dataset in the visible and infrared part of electromagnetic spectrum. Obtained results show that features from unmanned aerial vehicles are well separated from the other objects in image.