UTILITY OF ARTIFICIAL NEURAL NETWORK IN DETECTING URBAN ENVIRONMENT EXPANSION (A CASE STUDY: BABOLSAR, IRAN) Sedigheh Lotfi Dept. of Geography, University of Mazandaran: Babolsar, Iran E-mail: sedlotfi@yahoo.com


A new method based on an artificial neural network (ANN) is used to detect changes of the urban area of Babolsar since 1966 till 1994. The expanded urban area was depicted in aerial photographs of two different period. The research method uses two pairs of 23×23 centimetre aerial photographs of Iranian region acquired on different dates as input and supervises the ANN to classify the image data into «from-to» classes. Stereoscopic analysis was applied to extract the salient features and to reduce the dimensionality of the input data prior to the ANN-based change detection. Using Levenburg-Mar-quart (Hagan, Menhaj, 1994) algorithm, the three-layered backpropagation network was able to identify the changes of interest with an overall accuracy of 92.3 %. Compared to a conventional change detection algorithm, such as the post-classification, the results