Weak Target Detection In Sea Clutter

With the development of radar technology and the urgent requirements forsurface monitoring in many segments of society, weak target detection with highresolution and low grazing angle radar has aroused extensive concern by scholars inrecent years. Target detection in sea clutter can be equivalent to pattern classificationof binary signals. In technological approaches, the process of pattern classification isthe conversion of pattern space to type space through feature space. Thetransformation from feature space to type space is realized by classifier. The statistical,fractal, chaos, time-frequency distribution and FRFT domain fractal characteristics areanalyzed in this paper. Several characteristics which can be used as featurevectors areselected. In terms of engineering practice, the target correct detection probability andcomputational complexity are used as measurableindicators. The shapeparameter of Kdistribution, de-correlative time and spatial fractal character differences in FRFTdomain are finally selected as feature vectors. Support vector machine is used asclassifier to realize pattern classification in feature space. IPIX real-life sea clutterwas used for verification, and the results suggested that the proposed algorithm candetect weak target in sea clutter effectively. Besides, the proposed algorithmperformed a good behavior in the classification of primary and secondary target units.In this paper, the shape parameter of K distribution and de-correlative time areproposed as feature vector. Meanwhile, the fractal characteristic in FRFT domain ismodified. The optimization of feature space guarantees the effectiveness andpracticality of the detection algorithm, which has certain theoretical significance andpractical value.