Multi-Focus Plenoptic Images Dataset

Published: 09-02-2018| Version 1 | DOI: 10.17632/t6czryg5nw.1
Luca Palmieri


The Dataset consists in Plenoptic 2.0 Images - both real and synthetic ones. Real images were taken under controlled conditions using Raytrix Cameras, R29 and R42. Synthetic images are generated with a Blender Plug-in to emulate the plenoptic camera's behaviour. Every set of images comes with a calibration file (.xml file that contains the parameters). ___________ Real Images In the case of real images, the calibration file is inside the same folder (the same file applies for all images within that folder). For each scene is present also: - Raw image (black and white and not yet corrected) - Processed image (de-bayered and de-mosaiced image with colors) - Depth image (depth computed with our algorithm (see github code). It may be scaled for visualization purposes) ______________ Synthetic Images In the case of synthetic scene, each folder contains: - Rendered image - All-in-focus image - Rendered depth - All-in-focus depth - The depths estimated with different similarity measures (SAD, CENSUS, GRAD, NCC, SSD) - a .zip file that contains all micro-images rendered and the scene.json file with the relative informations. - the relative .xml calibration file (to emulate real images and their behaviour) Samples of how to use the image and a python library to work with such images is also provided on the github page (PlenopticToolbox2.0)