The tradition methods for estimating rolling period based on self correlation and spectrum analysis often gain false period. In allusion to this problem, a new method based on the nonparametric rank variance test and empirical mode decomposition is proposed. First, the trend character of RCS sequence is extracted by the mathematical morphological closed operation. Then, the extracted RCS sequence is operated by the empirical mode decomposition. And, the periods of each intrinsic mode functions(IMF) are gained by using the spectrum analysis Finally，the rolling period can be estimated by using the nonparametric statistics theory. The experimental results for the emulational and measured data show that the proposed method can solve the false period problem effectively and improve the performance of the estimation.

%U http://www.yhxb.org.cn/EN/10.3873/j.issn.1000-1328.2014.03.015