压缩超级像素遗传算法重现离轴电子全息图  

Reconstruction of Off-Axis Electron Hologram by Modified Genetic Algorithm with Compressed Supperpixel

作　　者：陈静[1] 孙琦[1] 顾昌鑫[1] 

机构地区：[1]复旦大学材料科学系,上海200433

出　　处：《真空科学与技术学报》2013年第1期24-29,共6页Chinese Journal of Vacuum Science and Technology

摘　　要：对重现离轴电子全息图的遗传算法进行了研究,讨论了参考波波矢和超级像素尺寸对重现结果的影响规律,并在此基础上对遗传算法进行了改进。遗传算法利用了全息图的全部信息因而得到较好的重现效果,与传统的滤波重现相比,该方法重现的振幅和相位误差分别下降了69%和64%。讨论结果表明,超级像素中必需包含一到两个周期的参考波以保证较好的重现效果,而参考波的取向对重现结果没有显著的影响;在垂直于参考波波矢方向上压缩超级像素的尺寸,可以进一步提高全息图的重现精度。本文提出了压缩超级像素遗传算法,与常规超级像素遗传算法相比,该方法重现的振幅和相位误差分别减小了15%和7%。We addressed the reconstruction of the off-axis electron hologram with the genetic algorithm.The influence of the reference wave and superpixel size on the reconstruction was evaluated,and a modified genetic algorithm with compressed superpixel（GA-CPS） was developed.When it comes to the reconstruction of the off-axis e-hologram,the newly-developed genetic algorithm takes the total information of the hologram into consideration,and outperforms the conventional Fourier transformation.Moreover,the reconstruction errors of the amplitude and phase were reduced by 69% and 64%,respectively.We found that a good reconstruction requires at least one cycle and no more than two cycles of the reference wave,and little depends on the orientation of the reference wave vector.Besides,the reconstruction errors of the amplitude and phase by GA-CPS algorithm were found to be 15% and 7% lower than those by the conventional genetic algorithm,respectively.

关 键 词：离轴电子全息 傅里叶变换 遗传算法 压缩超级像素遗传算法 

分 类 号：O242.23[理学—计算数学] TP18[理学—数学]