Meng Fanbo, Zhang Da, Wu Xizeng, Liu Hong
Center for Bioengineering and School of Electrical and Computer Engineering, University of Oklahoma, Norman, OK 73019, USA.
Opt Commun. 2009 Aug 15;282(16):3392-3396. doi: 10.1016/j.optcom.2009.05.027.
Previous studies have shown that iterative in-line x-ray phase retrieval algorithms may have higher precision than direct retrieval algorithms. This communication compares three iterative phase retrieval algorithms in terms of accuracy and efficiency using computer simulations. We found the Fourier transformation based algorithm (FT) is of the fastest convergence, while the Poisson-solver based algorithm (PS) has higher precision. The traditional Gerchberg-Saxton algorithm (GS) is very slow and sometimes does not converge in our tests. Then a mixed FT-PS algorithm is presented to achieve both high efficiency and high accuracy. The mixed algorithm is tested using simulated images with different noise level and experimentally obtained images of a piece of chicken breast muscle.
先前的研究表明，迭代式在线X射线相位恢复算法可能比直接恢复算法具有更高的精度。本通讯使用计算机模拟，在准确性和效率方面比较了三种迭代相位恢复算法。我们发现基于傅里叶变换的算法（FT）收敛最快，而基于泊松求解器的算法（PS）具有更高的精度。传统的格尔奇贝格-萨克斯顿算法（GS）非常缓慢，在我们的测试中有时无法收敛。然后提出了一种混合FT-PS算法，以实现高效率和高精度。使用具有不同噪声水平的模拟图像以及实验获得的一块鸡胸肌肉的图像对混合算法进行了测试。
