Shape Recognition in Drone Images Using Simplified Fuzzy Indexing Tables
Balázs Tusor, Ondrej Takáč, András Molnár, Štefan Gubo, Annamária R. Várkonyi-Kóczy
Abstract
In recent years, aerial drone photography has been rising in popularity not only as a civilian hobby and a cinematographic technique, but also in various engineering tasks as well. Shape recognition is a useful tool for many engineering tasks, such as structure inspection, traffic observation, etc. In this paper, a new shape recognition technique is proposed for recognizing shapes by their contours that can be used to quickly process images taken by drones. The image is first segmented into regions, then the contours are analyzed, and shape descriptors are built, which are used as inputs for the classifier. For pattern recognition, Simplified Sequential Fuzzy Indexing Tables are used in order to provide a very fast and robust inference.