Anderson-Darling检验在杂波分布辨别中的应用

Anderson-Darling test for clutter distribution identification

提出一种基于拟合优度检验的雷达杂波分布检验算法,采用Anderson-Darling检验代替传统假设检验,通过计算比较杂波样本的Anderson-Darling距离和备择分布簇临界值的大小来实现杂波分布辨别。研究中采用Monte-Carlo方法计算得到常用杂波分布簇在不同置信度水平和不同杂波样本长度下的临界值,并通过检验效能比较实验来验证本文算法的准确性,而实测海杂波和合成孔径雷达数据的杂波辨别实验也验证了该算法在现实应用中的有效性。

A novel method for radar clutter distribution identification is proposed, which makes use of Anderson-Darling goodness-of-fit （GoF） test to replace conventional hypothesis test. The judgment whether the clutter comes from certain specific distribution is made by comparing the Anderson-Darling statistic with the critical values of corresponding distribution family, which is derived using Monte Carlo computation for different significance levels and clutter serial lengths. The accuracy of the proposed method is presented in the power-of-test comparisons. Furthermore, the distribution identifications on sea clutter and SAR image were carried out to validate the proposed method.