Sea-surface Targets Detection Based On Strong Sea Clutter Background

The sea-surface floating small target detection has always been the focus of military scientific research.High-resolution radar not only needs to face the complex sea clutter characteristics,but also needs to overcome the sea-surface floating small targets echo characteristics.The detection method based on features has nice detection performance on the sea-surface floating small targets detection.Based on the joint multi-feature of distinguishable sea clutter and target echo,the detector is designed to improve the detection performance of he sea-surface floating small targets detection.The research contents are as follows:(1)The radar target detection problem is regarded as a two-classification problem,and the K nearest neighbor classification algorithm is used in machine learning to detect the seasurface floating small targets.Six existing features used to distinguish sea clutter and target are selected to form the joint feature vector.The existing six features are calculated the feature contribution rate by PCA method.An improved target detector is proposed by means of feature weighting.The experimental results show that this method can improve the detection performance and control the the false alarm rate of the detector.(2)The detection problem of sea-surface floating small targets is regarded as an anomaly detection problem.And the method is proposed on the basis of isolated forest anomaly detection.The existing six features data are uesd to construct the isolated forest.And anomaly scores is used to judge whether the target is included in the sample to be tested.The order of anomaly scores and the requirements of radar target for CFAR forms together the final target detector.Experimental results show that the detection performance can be improved by using isolated forest anomaly detector.(3)Based on the existing features,the new features for distinguishing clutter from target are proposed.It is combined with six existing features to form eight feature joint vectors.Based on the principal component analysis anomaly detection,a sea-surface floating small targets detection method is proposed.The method takes advantage of the principal component analysis fundamental to construct the target detector.According to error value judgment,the sample is judged as the target if error value is greater than the giving CFAR.Experimental results show that this method can improve the detection performance.