实现有效的图像去模糊对提高基于球面光学的计算成像系统性能具有极为重要的意义。分析了球面光学系统中的成像模糊模型，并介绍了基于维纳去卷积的图像去模糊算法。针对基于维纳去卷积的图像去模糊需要准确地估计模糊图像信噪比的问题，提出了一种新的基于图像去噪的图像信噪比自适应估计算法。分别使用Zemax光学仿真软件获得的图像和研制的基于球面光学的计算成像系统原理样机获得的图像开展实验研究，结果表明提出的算法能准确地估计出模糊图像的噪声方差和信噪比，利用估计得到的信噪比，使用维纳去卷积能得到比较理想的图像去模糊结果。因此，结合提出的方法，基于球面光学的计算成像系统能得到清晰的高分辨率图像。

It is significant to realize effective image deblurring for improving the performance of the computational imaging system based on spherical optics. The image blurring model in the spherical optics is analyzed, and the image deblurring algorithm based on Wiener deconvolution is introduced. To deal with the problem that the signal-noise ratio (SNR) should be estimated accurately in the image deblurring based on Wiener deconvolution, a novel adaptive SNR estimation algorithm based on image denoising is proposed. The experiments are performed using the images acquired by Zemax software and the implemented prototype of the computational imaging system based on spherical optics. The results show that the noise variance and SNR can be estimated with high accuracy by using the proposed algorithm, and good image deblurring results can be achieved using Wiener deconvolution with the adaptively estimated SNR, so the clear and high resolution images can be acquired by the computational imaging system based on spherical optics after integrating the work presented.