Small Target Detection on Sea Surface based on Feature-Weighted Support Vector Machine

Because the sea clutter is greatly affected by the complex and changeable environment, it is difficult to describe its characteristics through a single amplitude statistical model. In addition, the great variety and the small radar cross section of the target on the sea surface increase the detection difficulty. This paper proposes a feature-weighted SVM algorithm, which regards the detection of small targets on the sea surface as a binary classification problem. The algorithm extracts six features from different domains of the sea clutter, calculates the weight of the features according to the contribution of the features and constructs the feature space for detection with weighted feature vectors. Experiments on IPIX data sets show that the detection method improves the detection accuracy compared with the traditional algorithm and it is helpful for sea surface monitoring.