Detection of anthropogenic objects based on the spatial characteristics of their contour in aerial image

Abstract

In this paper, an analysis is performed for the spatial characteristics of the contours of aerial images that are homogeneous in brightness to determine the parameters that make it possible to identify anthropogenic objects with a higher probability. A complex criterion is proposed for the regions, which homogenous in brightness and an algorithm for detecting anthropogenic objects, which takes into account its area, the ratio of the total length of long contour fragments to the total length of the contour, the concentration of corners and endpoints of the contour. Also, when deciding on anthropogenicity, the particular characteristics of the contour of a homogeneous segment are used, depending on the number of long fragments of the contour. Using the proposed criterion, an algorithm of searching anthropogenic objects based on the analysis of contour elements (its shape, size and concentration) is developed. The proposed algorithm allows reducing the probability of skipping the anthropogenic object in comparison with the spatial anomaly search algorithm.