Auto-Focusing in Digital Holographic Microscopy Using Differential Cosine Score

Abstract

Accurately estimating the reconstruction distance of samples is very important for digital holographic microscopy, various auto-focusing algorithms have been proposed. These algorithms determine focus position by finding the global peak or valley of the evaluation curve. However, the discrete step size is small in precise autofocus process, which causes that the function values around peak or valley are not much different. As a result, the sensitivity of the peak and valley values of the evaluation curve is low, and the focusing accuracy is also susceptible to noise. In this paper, an autofocusing method using differential cosine score (DCS) is proposed to improve the accuracy and sensitivity of evaluation curve. In this method, two virtual detectors are placed on both sides of the real detector at the same distance. Two ordinary focusing curves and one DCS curve can be obtained by using the cosine score function. Therefore, the focal plane can be automatically identified by searching the zero-crossing of DCS curve, which avoids the problem of low sensitivity of peak and valley values by traditional methods. We illustrate the validity of the proposed method with measurements on USAF resolution target, blood smear and Hela cell.