Method of multilayer object sectioning based on a light scattering model

Abstract

We discuss a problem of reconstructing (sectioning) multilayer object images in observed images obtained by focusing the imaging system on each layer and containing spurious blurry images of neighboring layers. The blurring model used describes a physical process of incoherent light scattering in the Fresnel approximation with a priori unknown parameters of the point spread function. We propose a method of "Boundary separation" of sectioning, which combines the use of a physical blur model with modern methods of blur estimating and edge detection. The results of testing the "Boundary separation" method on the data of physical experiments with different-scale model multilayer objects are analyzed and compared with the existing methods for solving the optical sectioning problem. It is concluded that the method is most effective on multilayer objects with clearly defined boundaries, on which the method has demonstrated almost complete restoration of the desired layers.