The root mean square of mid-spatial-frequency wavefront (PSD1) is a key parameter used to evaluate the quality of mid-spatial-frequency wavefront of optical components. Before numerical calculation, the wavefront data should be filtered in frequency domain. Because the filtered data is easy to lead to data truncation in and outside the region to be measured, large edge high-frequency error and Gibbs noise are introduced, which seriously affects the accuracy of calculation. In order to reduce the impact of truncation, this paper optimally fills the truncated data to reduce the spatial frequency mutation inside and outside the truncated region. A four-way extended average algorithm is proposed by comparing the existing spatial preprocessing methods. Experimental results show that the proposed method can well restore the mid-frequency segment surface shape of optical elements and significantly improve the PSD1 measurement accuracy, and the mean error of the proposed method is less than 5% compared with the standard value.