The subaperture stitching algorithm as a key step of annular subaperture interferometric testing (ASIT) is used to sew all subaperture data together to get the full aperture information. The core part of stitching algorithm is that using the stitching model to eliminate the relative misalignment errors between the adjacent subapertures to realize the stitching treatment. Obviously, the stitching accuracy is mainly determined by the performance of the stitching model. Different stitching model for existing annular stitching algorithms are briefly discussed in this paper. Based on Zernike annular polynomials (ZAP), a more accurate and stabile stitching model for ASIT is proposed. By combining standard deviation and correlation coefficient of overlapping area, we can comprehensively evaluate the stitching performance. In addition, the fitting performance of the ZAP is evaluated in some pertinent cases: different obscuration ratios and noises. By comparing the results of proposed method with that of the classical auto-collimation method, it demonstrates that the accuracy and practicability of the proposed stitching model.