In digital holographic microscopy, one often obtains an in-focus image of the sample by applying a focus metric to a stack of numerical reconstructions. We present an alternative approach using a deep convolutional neural network.
Digital holography is a well-known technique for both sensing and displaying real-world three-dimensional objects.
Compression of digital holograms has been studied extensively, and the errors introduced by lossy compression are
routinely evaluated in a reconstruction domain. Mean-square error predominates in the evaluation of reconstruction
quality. However, it is not known how well this metric corresponds to what a viewer would regard as perceived error,
nor how consistently it functions across different holograms and different viewers. In this study, we evaluate how each of
seventeen viewers compared the visual quality of compressed and uncompressed holograms' reconstructions. Holograms
from five different three-dimensional objects were used in the study, captured using a phase-shift digital holography
setup. We applied two different lossy compression techniques to the complex-valued hologram pixels: uniform
quantization, and removal and quantization of the Fourier coefficients, and used seven different compression levels with
each.
KEYWORDS: Digital holography, Holograms, Image segmentation, Reconstruction algorithms, 3D image processing, Cameras, Sensors, 3D image reconstruction, 3D metrology, 3D displays
Digital holography allows one to sense and reconstruct the amplitude and phase of a wavefront reflected from or
transmitted through a real-world three-dimensional (3D) object. However, some combinations of hologram capture setup
and 3D object pose problems for the reliable reconstruction of quantitative phase information. In particular, these are
cases where the twin image or noise corrupts the reconstructed phase. In such cases it is usual that only amplitude is
reconstructed and used as the basis for metrology. A focus criterion is often applied to this reconstructed amplitude to
extract depth information from the sensed 3D scene. In this paper we present an alternative technique based on applying
conventional stereo computer vision algorithms to amplitude reconstructions. In the technique, two perspectives are
reconstructed from a single hologram, and the stereo disparity between the pair is used to infer depth information for
different regions in the field of view. Such an approach has inherent simplifications in digital holography as the epipolar
geometry is known a priori. We show the effectiveness of the technique using digital holograms of real-world 3D
objects. We discuss extensions to multi-view algorithms, the effect of speckle, and sensitivity to the depth of field of
reconstructions.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.