开发了基于Geant4的Z箍缩中子编码成像系统模拟平台,实现聚变中子编码成像诊断系统各关键部件的完整模拟.获得了低中子产额(约1010量级)下,中子经编码孔编码后在闪烁体阵列中形成的发光分布图像.利用维纳滤波、Richardson-Lucy(RL)及遗传算法(GA)对低中子产额下获得的极低信噪比图像进行重建,并对信噪比、中子产额及重建效果进行了对比研究,结果表明:遗传算法对低信噪比中子编码图像的重建具有很强的鲁棒性;中子编码图像的信噪比与遗传算法重建结果的准确性呈正比.%The model of Z-pinch driven fusion imaging diagnosis system was set up by a Monte Carlo code based on the Geant4 simulation toolkit. All physical processes that the reality involves are taken into consideration in simulation. The light image of low neutron yield (about 1010) pill was obtained. Three types of image reconstruction algorithm, i. e. Richardson-Lucy, Wiener filtering and genetic algorithm were employed to reconstruct the neutron image with a low signal to noise ratio (SNR) and yield. The effects of neutron yields and the SNR on reconstruction performance were discussed. The results show that genetic algorithm is very robust for reconstructing neutron images with a low SNR. And the index of reconstruction performance and the image correlation coefficient using genetic algorithm, are proportional to the SNR of the neutron coded image.
展开▼
