In order to improve the wave-front aberration reconstruction accuracy of human eye in adaptive optics system, as well as the accuracy of point spread function, an algorithm of Zernike model based on partial ridge estimation for wave-front aberration reconstruction was proposed. First, based on the Zernike model, the algorithm partial ridge estimation was bought into wave-front aberration reconstruction. Second, in the sense of mean squared errors, the parameter of ridge estimation was ensured to improve the singularity of least square and restrain the enlarger of wave-front measure error. Finally, the error produced by two different kinds of algorithm was discussed with the influence of wave-front slope error. The results show that the partial ridge estimation can decrease the mean squared errors of the modal coefficient effectively. Compared with the singular value decomposition, the mean squared error is reduced by 8%~10%. With the increasing of wave-front measurement error, the accuracy of the partial ridge estimation is much better than the least square estimate, which is useful for studying the image restoration based on wave-front detection in the late of adaptive optics.