中文摘要：

在实际显微系统中，由于样本的折射率和透镜折射率不匹配，致使不同深度的PSF是不一样的．为了实现三维显微图像的复原，提出了基于三维高斯PSF的复原算法，将Hopfield神经网络用于三维图像序列的复原中，实验证明连续Hopfield网络能够恢复深度变化图像模型的模糊图像．

英文摘要：

A full-parallel Hopfield neural network is presented based on a new mod al （3D） Gaussian point of the diffraction of the specimen and the one of spread function （PSF） to restore t microscope＇s objective and the di the microscope, it leads to the diff he microscopic op fference between el of the three dimension- tical slices. As the result the refraction rate in the erent PSF among the depth-variant plane in the practical microscopic system. In order to restore the 3D microscopic image, a new restoring algorithm is proposed based on the 3D Gauss PSF. It is the first time that the Hopfield network is used in 3D image restoration, and the performance shows that the continuous Hopfield network can restore the blurred image of the depth-variant image model.