基于深度变化成像模型的调整EM算法

Regularized EM image estimation based on depth-variant imaging model

在实际成像中,通常样本中的物质是变化的,故样本中不同位置的折射率不一样。由于三维样本的折射率与物镜所浸物质的折射率的不匹配,导致不同深度的点扩展函数可能不同。在此深度变化成像模型基础上应用最大期望(EM)复原算法能够提高图像清晰度,尤其是深度方向,但会丢失图像的一些微弱细节且出现一些孤立亮点,因此将调整EM算法运用到基于三维显微光学切片中成像随深度变化的图像模型上,此二者结合后的新算法可以避免上述缺点,较好地恢复图像微弱细节。

In practical imaging, because the substance of a specimen varies spatially, the refractive indexes in different depth are different. A large mismatch of the refractive index of 3D specimen and immersion medium leads to different PSFs in different depths. Using Expectation Maximization （EM） algorithm based on the depth-variant imaging model can improve image resolution, especially in depth, but it would result in loosing dim detail and enhancing very bright isolated spots. A regularized EM algorithm was used to avoid disadvantages and recover the detail of image in the depth-variant imaging model in threedimensional optical sectioning microscopy.