The empirical mode decomposition (EMD) proposed by Huang et al. in 1998 shows remarkably effective in analyzing nonlinear signals. It adaptively represents nonstationary signals as sums of zero-mean amplitude modulation-frequency modulation (AM-FM) components by iteratively conducting the sifting process. How to determine the boundary conditions of the cubic spline when constructing the envelopes of data is the critical issue of the sifting process. A simple bound hit process technique is presented in this paper which constructs two periodic series from the original data by even and odd extension and then builds the envelopes using cubic spline with periodic boundary condition. The EMD is conducted fluently without any assumptions of the processed data by this approach. An example is presented to pick out the weak modulation of internal waves from an Envisat ASAR image by EMD with the boundary process technique