【摘要】 本文以ＣＣＤ图象为技术背景，指出ＣＣＤ图象噪声主要将以与信号有关的泊松噪声形式存在。根据图象形成的模型及图象统计特性，提出了最大似然（ＭＩ）法，最大后验概率（ＭＡＰ）法及最大熵（ＭＥ）复原方法。针对ＭＡＰ法提出了具体算法和参数估计方法。并提出应用图象分割法来提高计算机复原速度，节省存储空间。更多还原

【Abstract】 In this paper it is pointed out that the noise in a CCD image will be mainly the Poisson noise, which is the signal-dependent noise, with the technological background of CCD images. Using statistics and model of the image formation, we present three different nonlinear restoration methods: Maximum Likelihood (ML), Maximum a Posteriori (MAP), Maximum Entropy (ME). We derive the algorithm and parameters estimate method for the MAP restoration, and propose the image sectioning to speed up the restoration and save memory space.更多还原