Conference Publication Details
Mandatory Fields
Fan X.;Healy J.;Guanshen Y.;Hennelly B.
Proceedings of SPIE - The International Society for Optical Engineering
Sparsity metrics for autofocus in digital holographic microscopy
4 ()
Optional Fields
autofocus biological cell Digital holographic microscopy sparsity metrics
© 2016 SPIE. Digital holographic microscopy is an optic-electronic technique that enables the numerical reconstruction of the complex wave-field reflected from, or transmitted through, a target. Together with phase unwrapping, this method permits a height profile, a thickness profile, and/or a refractive index profile, to be extracted, in addition to the reconstruction of the image intensity. Digital holographic microscopy is unlike classical imaging systems in that one can obtain the focused image without situating the camera in the focal plane; indeed, it is possible to recover the complex wave-field at any distance from the camera plane. In order to reconstruct the image, the captured interference pattern is first processed to remove the virtual image and DC component, and then back-propagated using a numerical implementation of the Fresnel transform. A necessary input parameter to this algorithm is the distance from the camera to the image plane, which may be measured independently, estimated by eye following reconstruction at multiple distances, or estimated automatically using a focus metric. Autofocus algorithms are commonly used in microscopy in order to estimate the depth at which the image comes into focus by manually adjusting the microscope stage; in digital holographic microscopy the hologram can be reconstructed at multiple depths, and the autofocus metric can be evaluated for each reconstructed image intensity. In this paper, fifteen sparsity metrics are investigated as potential focus metrics for digital holographic microscopy, whereby the metrics are applied to a series of reconstructed intensities. These metrics are tested on the hologram of a biological cell. The results demonstrate that many of the metrics produce similar profiles, and groupings of the metrics are proposed.
Grant Details