A Hybrid Features–Based Target Detection Scheme for Maritime Radar

The popular method of automatic target detection used for most maritime radar systems is constant false alarm rate. However, this technique seems insufficient to differentiate a real target from others. To overcome such a deficiency, a hybrid features–based target detection methodology is developed to distinguish between target and non–target objects for maritime radar. This approach combines echo signal detection based on constant false alarm rate and support vector machine classification using hybrid features extracted from objects resulting from connected–component labeling. The efficacy of the proposed approach is verified with datasets recorded from a real maritime radar system.