An Algorithm for Detecting the Location of Rodent-Made Holes through Aerial Filming by Drones

Order Rodentia is a massive group of animals. In case the population of which is not controlled, their sprawl and propagation will impose severe problems for humankind. One way to control and fight the growth of the Rodentia population is by poisoning their burrows, which requires finding the precise location of holes made. The present research provides a fast, intelligent algorithm to locate holes of rodent burrows. A quadcopter with a 1-m/s propeller speed at a constant 1-m height from the ground was used for aerial filming. Filming rate was 20 frames per second. Movies prepared by the camera were transmitted to a computer via wireless communications and converted into images by Aoao software. From images, red, green, blue, and gray surfaces, as well as co-occurrence matrices, were extracted, from which, in turn, 40 color and 44 texture features were extracted. Once top features, most of which related to the texture of images, were selected, they were used to train and implement the classifier of a support vector machine (SVW) with the radial base kernel. Classification accuracy decreased with an increase in Kernel breadth so that maximum classification accuracy was obtained with Kernel breadth of 0.1. Results showed 96.1% accuracy for the proposed method in locating the burrow holes.