DarkFocus: numerical autofocusing in digital in-line holographic microscopy using variance of computational dark-field gradient

Abstract

We report on a novel computational technique for automatic numerical refocusing in digital in-line holographic microscopy. It is based on the adaptive filtering of the recorded on-axis hologram to eliminate its background term and extract interference intensity-component connected with light scattered on the sample (interference fringes). Numerical propagation of such filtered hologram yields the computationally generated dark-field imaging coming from the amplitude part of the complex field. We propose a simple measure in the form of the variance of the dark-field gradient, which attains its maximum value in the focal planes for all types of objects (phase, amplitude and mixed phase-amplitude), as a robust indicator of overall sharpness and in-focus quality. Demonstrated novel autofocusing technique is positively validated using experimental data exhibiting significant variation of the magnification factor for both single and double focal plane scenarios. It compares favorably with other well-established automatic numerical refocusing techniques based on the high-pass filtered complex amplitude and edge sparsity, mainly in terms of higher axial resolution and better robustness to hologram low signal-to-noise ratio.