Abstract:

This paper proposes a new approach to ship detection based on polarimetric whitening filter (PWF). The Markov random field (MRF) classification problem based on Wishart distribution is introduced, and similarity parameters and distance function are employed to construct the smooth item for using polarization information effectively. The Fisher criterion is used to select the parameters when using Nash equilibrium to optimize the classification result. For the final detector, the detection function of data items corresponding to the energy function is constructed by solving an optimization problem. For this problem, PWF structure is used to construct a kernel function of differences. At the same time, the classification result and fuzzy set are applied to construct the detection function of the smooth item. Then a new detector is proposed. Using polarimetric synthetic apertue radar (PolSAR) data, the effectiveness of the proposed algorithm is demonstrated and is compared with the existing algorithms, and the experiment verifies the high detection rate of the proposed method with the good continuity of target detection.