Validation of Ship Detection Targets from High Frequency Surface Wave Radar

Abstract

Due to the heterogeneous signals received by the HFSWR, there are signals that are irrelevant and redundant. Further, the CFAR used to detect the targets was validated through visual interpretation using the AIS data as the reference. This study performed OBIA to classify true targets and possible clutters which were validated using the AIS data. The classification performed the SVM based on Gaussian RBF. Overall accuracy from error matrix based on test and training mask acquired 92%, 91% and 90% respectively. Hence, the classification methodology produced acceptable accuracy. Pre-processing of the dataset may be required producing better accuracy.