基于光学衍射的纳米尺度深度重建

Nano scale depth reconstruction based on optical diffraction

传统的离焦深度恢复方法(depth from defocus,DFD)在计算离焦图像的模糊程度时,均是基于几何光学成像原理,并没有考虑光波衍射对图像模糊的影响.然而,衍射现象是波的基本特性,以针孔成像为基础的光学成像系统中必然存在光波衍射.本文基于光学衍射和图像模糊机理,提出了一种精确的单目视觉全局景物3D深度信息获取方法.首先,详细分析了光学成像系统中的菲涅耳衍射机理,建立了衍射模糊与景物深度之间的关系曲线;然后,采用曲线拟合的方式建立了衍射模糊的模糊程度(点扩散函数扩散程度)与景物深度的数学关系模型,结合热辐射和图像相对模糊原理,构建了衍射模糊热辐射方程组,并把衍射模糊深度获取问题转换成一个深度信息动态优化问题.最后,使用标准的纳米栅格模板验证了衍射模糊深度重建算法在纳米尺度相对深度和绝对深度重建中的有效性和精确性.

Traditional depth from defocus（DFD） is based on the imaging principle in geometric optics to calculate blurring degree, influence of optical diffraction has never been considered. However, diffraction is a basic property of all kinds of waves, it must happen in an optical imaging system based on pinhole imaging principle. In this paper, a high resolution DFD method with a single vision and optical diffraction is given. Firstly, the basic principle of Fresnel diffraction in an optical system is analyzed, and the relationship curve between Fresnel diffraction and depth information is constructed. Secondly, a defocus imaging model with optical diffraction is supposed through curve fitting, and with respect to relative blurring and heat diffusion, a new DFD method with optical diffraction and two defocused images is proposed. Finally, a standard nano grid is used to validate the proposed DFD method on nana scale. The results show that the new algorithm, compared with traditional DFD algorithms, is a more effective method to reconstruct depth information on nano scale.