A Novel Method for Road Extraction from Satellite Images

Extended Kalman filter (EKF) has previously been used to extract road maps in satellite images. The extended Kalman filter in general is not an optimal estimator, if the measurement and the state transition model are both linear. In addition, if the initial estimate of the state is wrong, or if the process is modeled incorrectly, the filter may quickly diverge, owing to its linearization. In our new approach, we have combined Unscented Kalman Filter with a special Particle Filter (LLPF) in order to regain the trace of the road beyond obstacles, as well as to find and follow different road branches after reaching to a road junction. In this approach, first, EKF module starts tracing the road using the initial state and the initial profile cluster. While progressing along the road path, the profile clusters are updated, and new appropriate clusters are added as road intensities and/or widths change. Then passed it to LLPF(Local Linearization Particle Filter), which tries to find the continuation of road after a possible obstacle or to identify all possible road branches that might exist on the other side of a road junction. For further improvement, we have modified the procedure for obtaining the measurements by decoupling this process from the current state prediction of the filter.