Small target detection in sea clutter based on linear prediction

To detect small target in sea clutter, a method based on local backscattering amplitude prediction error is proposed in the paper. This approach builds on linear prediction: calculating the mean of the fourth power of prediction error of backscattering amplitude. By analyzing the IPIX radar datasets, it is found that the prediction error´s mean and variance of the range bin containing a small target is much smaller than that of other range bins containing only sea clutter. The range bin, in which a small target is contained, gets smaller prediction error than those bins in which only pure sea clutter is occupied. Upon this point, we detect the small target in sea clutter with small amount of priori knowledge. The experimental results show that the proposed method can effectively detect small target in sea clutter with better performance.