System matrix generation for 3D tomographic reconstructions using an optical design software

3D imaging of solid targets using plenoptic cameras has been extensively explored and optimized over the years. Unfortunately, these imaging techniques, typically making use of triangulation methods and contrast recognition for depth estimation, lose their validity when imaging translucent media. For these cases, tomographic reconstruction has been shown to be a promising avenue for recovering the 3D shape of a translucent volume. Apart from the tomographic reconstruction algorithm itself, the accuracy of a reconstructed solution depends on the set of measured projections and on the system matrix. A proper determination of the system matrix is key as its elements describe the weighted contribution of a voxel in object space to one in image space; however, computing the matrix elements can consist of an arduous task as it requires a priori knowledge of the imaging system and precise modeling of the physical properties of the acquisition process. In this work, we present how an optical design software can be used to generate such system matrices. Compared with approximation methods involving paraxial ray tracing, the proposed method offers the main advantage of real ray tracing, in which the computed weighted contributions intrinsically account for optical aberrations in the imaging system. Physical properties of light propagation within the translucent medium can also be taken into account when using non-sequential modes. Using a ray tracing software thus offers great flexibility in designing plenoptic imaging systems used with 3D tomographic reconstruction techniques.