二维最大熵成象初探

A Research for the Two-dimensional Maximum Entropy Image Reconstruction

最大熵成象法能从不完全和带噪数据成象,在许多领域有重要应用。其主要的差别在于.各种学科在观测方程中所用的影响函数不同。在最大熵目标函数中采用适当的影响函数,就能使这种成象法用于各种专门研究。本文详尽地论述了傅里叶数据形式下的成象过程。最大熵法是迭代型的,因而比直接法成象要多耗费。然而图象的质量和分辨力显著提高,迭代次数不过几次就能达到这个水平。这两种方法所得的图象在图3中作了比较。

Maximum Entropy Image Reconstruction has potential applicationsin many fields concerned with reconstruction from incomplete and noisy data.The essential difference between reconstruction in these varied disciplines is inthe form of the kernel in the measurement equation. Incorporation of the appro-priate kernel into the maximum entropy objective function a dapts the methodto new specialities. Implementation is described in detail for the Fourier datatype. The method is iterative and hence more costly than direct technique.However, a significant improvement in image quality and resolution is possiblewith only a few iterations. Reconstructions using the two methods are displayedin Fig. 3.