An image restoration method for ground-based adaptive optics telescope imaging was proposed in the case of simultaneous recording of short exposure images and wavefront sensor measurements. The proposed method was based on a zonal representation of the turbulence-degraded wavefront phase using bivariate simplex splines function, instead of the traditional Zernike modal decomposition; an average slopes measurement model for Shack-Hartmann wavefront sensor was then built based on the zonal representation method, which transformed the ill-posed wavefront reconstruction problem into a well-conditioned equality constrained least squares problem. The object image was finally obtained by non-blind Richardson -Lucy iterative deconvolution. Simulated experiments show that the proposed method performs superior image restoration results and noise suppression capability in different turbulence strength.