In Remote sensing systems one of the most important features needed are roads, which require automated procedures to rapidly identify them from high-resolution satellite imagery, Many approaches for automatic road extraction have appeared in literature [2][7][9], which vary due to the differences in their goals, available information, algorithms used and assumptions about roads. In this paper, we propose an approach for automatic road extraction by integrating region and edge information. The complimentary information of road segments obtained using Probabilistic SVM (PSVM) and road edges obtained using Dominant Singular Measure (DSM) are integrated using a modified Constraint Satisfaction Neural Network -Complementary Information Integration(CSNN-CII) [1] to improve the accuracy of the system. Results are shown on real-world images and quantitatively evaluated with manual hand-drawn road layouts. © 2009 Crown Copyright.