Mumford-Shah model based man-made objects detection from aerial images

Abstract

In this paper, a novel method for detecting man-made objects in aerial images is described. The method is based on a simplified Mumford-Shah model. It applies fractal error metric, developed by Cooper, et al [1] and additional constraint, a texture edge descriptor which is defined by DCT (Discrete Cosine Transform) coefficients on the image, to get a preferable segmentation. Man-made objects and natural areas are optimally differentiated by evolving the partial differential equation using this Mumford-Shah model. The method artfully avoids selecting a threshold to separate the fractal error image, since an improper threshold may result large segmentation errors. Experiments of the segmentation show that the proposed method is efficient. © Springer-Verlag Berlin Heidelberg 2005.