为提高图像盲复原处理效果，提出了经验法、拟合高斯点扩散函数法,以及符合Kolmogorov谱函数的初值选取方法等三种初值选取方法。引入泽尼克多项式参量化表示点扩散函数，应用极大似然迭代盲解卷积算法对模拟模糊图以及木星观测图进行了复原处理。计算结果表明，符合Kolmogorov谱函数分布的初值方法以及拟合高斯点扩散函数方法得到的图像复原结果较好。

To improve the quality of the image restored by blind deconvolution, three initial-value-choosing methods are proposed, namely the experimental method, the Gaussian point spread function fitting method and the Kolmogorov spectrum method. Zernike polynomials are introduced to parameterize the point spread function, and the maximum-likelihood iterative blind deconvolution algorithm is applied to restore the blurred SEASAT image and the observed image of Jupiter. Experimental results show that the images have better details when restored by applying the Kolmogorove spectrum method and the Gaussian point spread function method.