Usage of Fractal Dimension for Identification of Man-Made Objects

Automatic vectorization of aerospace images causes redundancy of node points of man-made objects' contours. Usage of Hough parametric space for reduction of approximation points, unlike the piecewise linear method, allows receiving more accurate representation of contours, but its computer realization is rather time-consuming. In the paper in order to reduce time and to increase efficiency of the process of vectorization of images of ultrahigh spatial resolution the possibility of usage of fractal dimension as an additional criterion for objects classification is investigated.