This work presents a method for simulating digital lensless holographic microscopy (DLHM) holograms using a physics-based image processing approach. While DLHM has gained significant attention in biology, biomedicine, and environmental monitoring, the current modeling of DLHM holograms has been limited, hindering potential applications, including learning-based solutions and generative model training. In this study, the DLHM propagation process is decomposed into the diffraction of a complex-valued spherical wavefront and the non-homogeneous magnification of the diffracted field that encodes the sample information, which accelerates and enhances the hologram simulation. The proposed model is validated by comparing simulated and experimental holograms of standard test targets under diverse imaging conditions. Comparative analyses are conducted against other DLHM hologram modeling methods, including direct Rayleigh-Sommerfeld diffraction, its convolutional implementation, and the Fresnel-Bluestein formalism. The proposed model is shown to outperform these methods in overall similarity to experimental recordings across a wide range of imaging conditions while maintaining computational efficiency. This DLHM hologram modeling approach provides researchers with a powerful tool for simulating trustable holograms. The model can be publicly accessed through the open-access repository https://github.com/mloper23/DLHM-model.
Read full abstract