Research on a Structure Analysis of Materials with a Maximum Entropy Restoration of Electron Microscope Images.

1.最大熵（ME）恢复程序的实现：由量子噪声的空间随机分布约束的ME图像恢复的计算机程序被开发并在PC上实现。该程序允许可视化的过程中的图像的变化，以及修改后的ME恢复和传统的kai平方约束方法之间的差异。这也使我们能够在傅立叶空间中的ME恢复的效果进行研究。2.弱相位对象的TEM图像的恢复：修改的ME方法被应用到大散焦的铁蛋白颗粒的TEM图像的反卷积。由于散焦图像中的对比度传递函数（CTF）严重振荡而被反向对比度模糊的铁蛋白颗粒的核心被恢复到具有正确对比度的原始尺寸。傅里叶分析表明，即使在CTF中的零附近，由于ME恢复的结果，散焦图像中的反向对比度的频率区域也会通过π的相移自动重建。3.晶体成像的实现：为了将ME修复应用于晶体标本，需要TEM成像的模拟，以从样品正下方的模拟复波函数计算图像强度，并将图像强度与观察到的图像进行比较。因此，图像模拟程序，用C++语言实现，并纳入ME恢复程序。模拟结果与标准程序计算结果吻合较好。4.高分辨率晶体图像的ME恢复方法的发展：现有的ME方法由于设计的原因，不能应用于晶体图像。当电子波函数是复数时，只处理真实的数。为了解决这一问题，将该算法扩展到处理复杂函数。

1.Implementation of Maximum Entropy (ME) Restoration Programs : Computer programs for the ME image restoration constrained by the spatially random distribution of quantum noise were developed and implemented on PCs. The programs allows visualization of changes in the image during the process of restorations and the difference between the modified ME restoration and conventional kai-squared constrained method. It also enables us to study the effect of the ME restoration in the Fourier space.2.Restoration of TEM Images of Weak-phase Objects : The modified ME method was applied to deconvolution of largely defocused TEM images of ferritin particles. The cores of ferritin particles, which were blurred with reversed contrast due to the severely oscillating contrast transfer function (CTF) in the defocused image, are recovered to the original size with the correct contrast. The Fourier analysis showed that the frequency regions of reversed contrast in the defocused image is automatically reconstructed by the phase shift of pi as a result of the ME restoration even around zeros in the CTF.3.Implementation of the Imaging of Crystals : For application of the ME restoration to crystal specimens, simulation of TEM imaging is required to calculate the image intensity from a simulated complex wave function just below the specimen and to compare the image intensity with the observed image. An image simulation program was, therefore, implemented with a C++ language and incorporated into ME restoration programs. The results of the simulation were in good agreement with those calculated with a standard program.4.Development of ME Restoration of High-resolution Crystal Images : Existing ME methods cannot be applied to crystal images since they are designed. to treat only real number while the electron wave function is complex. To solve this problem, the algorism was extended to deal with complex function.