An Efficient Method for Road Tracking from Satellite Images Using Hybrid Multi-Kernel Partial Least Square Analysis and Particle Filter

Abstract

The Road extraction from the remotely sensed imagery is highly realistic for the quick road updating in the Geographic Information System (GIS) data collection. The particle filter (PF) was earlier employed to track the road maps in satellite images. In our previous work, we have introduced an efficient Gauss–Hermite Kalman Filter with Locally Excitatory Globally Inhibitory Oscillator Networks (GHKF–LEGION)-based road extraction, even though it does not properly extract the road from the complex region. In order to recover the track of the road beyond obstacles, in this work, we proposed a novel hybrid multi-kernel partial least squares (PLS) with PF approach. Here, at first, we estimate the initial leader point of the road employing the K-means clustering technique. Subsequently, the PF traces a road till a stopping benchmark is satisfied. Thereafter, without finishing the process, the outcomes are furnished to the hybrid kernel PLS technique which attempts to locate the continuance of the road after several potential road blocks or to locate the entire feasible road branches which are on the other side of the road junction. The outcomes are offered for five satellite images. The experimental results show our proposed road tracking method is better compared to other existing works.