Robust Depth Estimation for Light Field Microscopy
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
DENOISING
DEPTH
GABRIELE
GCO_v3.0
GROUNDTRUTH
IMAGES
JEON2015
MATTING
MERGING
MGM2015
PARAMETERS
PRIORS
READ
SAVE
SFF2016
SUPERPIXELS
TAO2013
VISUALIZATION
readmeimages
vlfeat-0.9.20
LICENSE
Main.m
README.md
generateErrorResults.m
setupPaths.m

README.md

Robust Depth Estimation for Light Field Microscopy

This is a Matlab implementation of the procedure described in the paper:

Robust Depth Estimation for Light Field Microscopy, L. Palmieri, G. Scrofani, N. Incardona, G. Saavedra, M. Martínez-Corral and R. Koch, Sensors 2019, 19, 500.

Algorithm

Starting from a FiMic image, it creates a focal stack and separates the perspective views. Then two cost volumes are built, and refined using multi-scale and superpixels contributions. Finally, using energy minimization the two volumes are combined and the final depth map is extracted.

Some Results

Cotton fibers Cotton fibers Zebrafish Chip Solderings

Microscope Imagery

The FiMic microscope allows to capture light field in a single shot.

For further information regarding FiMic Fourier Integral Microscope, refer to the paper FIMic: design for ultimate 3D-integral microscopy of in-vivo biological samples

Implementation

About implementation, the code is mainly in Matlab and it makes use of some MEX files (C and C++) to speed up some computations. It uses other toolboxes, but they are already present in the github folders (see References for more details).

Dependencies

The dependencies are a C++ compiler for the MEX code (if you are using gcc, supported version is 4.7; however until 6.x is working and Matlab just gives a warning, from 7.0 it may cause some problems. If on Linux systems, use of alternatives can help about this read more here)

Main References

It takes inspiration and parts of code from other approaches, who have been cited in the paper and we would like to thank for sharing their code.

[1] JEON2015: Open Source Code -- Accurate Depth Map Estimation from a Lenslet Light Field Camera, Jeon et al., 2015

[2] MGM2015: Project page -- Open Source Code -- MGM: A Significantly More Global Matching for Stereovision, Facciolo et al., 2015

[3] SFF2016: Open Source Code -- Analysis of focus measure operators for shape-from-focus, Pertuz et al., 2013

[4] TAO2013: Project Page with Open Source Code -- Depth from Combining Defocus and Correspondence Using Light-Field Cameras, Tao et al., 2013

For further information, please contact me at: lpa@informatik.uni-kiel.de