Determination of number of termite mounds supported by computational vision

Abstract

The determination of the number of termite mounds in a pasture area is a factor that can determine the stage of degradation of the same. The termite count work has been done manually which becomes costly. Our work aims to present a comparison of image filtering techniques and the best time to collect aerial images with UAV (Unmanned Aerial Vehicles) in order to count the number of termite mounds by means of computer vision. Comparisons were made between the number of mounds counted in the image by a specialist with the number of mounds in the image after being submitted to computer vision techniques. It was identified that height is the main variable and that of the five techniques used, Gaussian and Bilateral presented the best results.