Fast Fourier Transform (FFT) is one of the most important interferogram analysis methods with the merits of single interferogram captured, low experimental environment requirements and better accuracy. FFT arithmetic can only process numeric discrete data and requires that the number of line and row pixels must be 2 to the power n. But unfortunately, interferograms captured are circular in general. The fringe extrapolation method has been proved to be very effective in avoiding Gibbs phenomenon and reducing phase evaluation errors. Gerchberg proposed a simple iterative algorithm to extrapolate the interferograms, but there is not a good evaluation criterion of the iterative times. In this paper, a method of exemplar-based image inpainting is proposed. First, the priority of each patch on the "fill front" should be calculated. We define its priority P(p) as the product of two terms p(p)=C(p)D(p).C(p) is the confidence term and D(p) is the data term. The patch with the highest priority is obtained. Then, the best exemplar patch corresponding to the patch with the highest priority is discovered in the exemplar region. And the relevant data is copied from the best exemplar patch to the unfilled parts of the patch with the highest priority. Finally, the confidence values of pixels which have been filled just now are updated. The whole region will be fully filled through iterating the above steps. Computer simulation and experiment make it clear that the proposed algorithm extrapolate the texture and structure information effectually.