The point spread function of atmospheric turbulence is space variant where the field-of-view of the ground-based optical telescope is greater than the isoplanatic angle. The effects of anisoplanatism in image degrade the image quality recovered by conventional image processing based on the imaging technique of the space-invariant point spread function. An approach is to partition the ansioplanatic image into sub-region images corresponding to the isoplanatic angle. The sub-region images are then reconstructed by the conjugate gradient blind deconvolution algorithm. To avoid ringing at the edges, a window function is used to smooth the image edges. After each sub-image is restored, the full image is assembled by the seamless split and overlapping. The experimental results suggest that the proposed approach is effective to improve the restored image quality from the anisoplanatic image degraded by atmospheric turbulence.