Abstract: A restoration approach of the circulation edge algorithm in frequency domain based on the Wiener filtering is proposed. The effective regularization expression and the estimation method of the signal to noise ratio(SNR) are presented. First, the original supervised image is extended with the reflection symmetry manner to a new supervised image (NSI). Then, the NSI is restored by the Wiener filtering algorithm, and the primary value sequence of the image restored by the NSI is considered as the restored image which is corresponding to the original supervised image. When the NSI is restored in frequency domain, the fast Fourier transform (FFT) technique need to be used and the NSI is extended periodically. Thus the gradient on the extended edges in the orthogonal orientation is zero, and the differential condition is satisfied with the edge smoothness. Therefore, the ringing ripples produced by the sharp changes of the extended edge′s gradient are suppressed. In the restoration process, the ratio of the noise power spectrum to the original image power spectrum is regularized to the function of the SNR of the original supervised image. The experimental results show that the restoration effect using this approach is better than the one using Wiener filtering for the supervised image having large gradient changes on the area near to its edges.