A New Neural Architecture for Detecting Urban Changes in Quickbird Imagery

High-resolution imagery presents a new challenge over other satellite systems in that a relatively large amount of data must be analyzed and corrected for registration and classification errors to identify the land cover changes related to urban development. To obtain the accuracies required by typical applications for wide areas, very extensive manual work is commonly required. In response, we have developed a new method for urban change detection that greatly reduces the human effort needed to analyze the high-resolution imagery. The technique consists in considering a neural network architecture able in parallel to exploit either multitemporal or multispectral satellite information. We found that this new technique is very accurate relative to the results yielded by an image processing approach based on careful visual inspection.