This PDF 1.4 document has been generated by Primoz Peterlin / GPL Ghostscript 9.10, and has been sent on pdf-archive.com on 05/11/2015 at 17:01, from IP address 192.93.x.x.
The current document download page has been viewed 734 times.
File size: 37.2 MB (162 pages).
Privacy: public file
A Full-Featured Open-Source Framework
for Image Processing
http://gmic.eu
November 2015
Context
Research in the field of image processing at the GREYC lab of ENSICAEN /
CNRS / University of Normandy (Caen).
⇒ Design of innovative algorithms to solve generic image processing problems
(denoising, enhancement, segmentation, feature detection,...).
Context
Frequent collaborations with companies / laboratories having specific images to
process.
...
⇒ Various image data coming from very diverse sensors.
Context
Image data are diverse: 2D, 2D+t, 3D, 3D+t, vector or matrix-valued pixels, float
values, ...
⇒ We stray far from usual 2D color pictures!
(a) I1 ∶ W × H → [0, 255]3
(b) I2 ∶ W × H × D → [0, 65535]32
(d) I4 ∶ W × H × T → [0, 4095]
(c) I3 ∶ W × H × T → [0, 4095]
Motivations
Needs for tools to visualize / explore data, convert image formats, apply classical
IP operators (filtering, geometric transformations, frequential analysis, ...) for very
generic image data, sometimes on thousands of images at the same time.
Typical “technical” question heard at the lab: “How may I easily convolve 500
volumetric images with 32 channels each by a 3d anisotropic gaussian kernel ?”
Motivations
⇒ Very few open-source tools exist for these tasks. They tend to be either:
▸
Easy to use, but not generic enough for our image data (ImageMagick,
GraphicsMagick, ...).
▸
Or very flexible, but reserved for experienced programmers (require the writing of code,
using specialized external libraries).
We did like others: Since 199x, we have developed such generic libraries for
image processing: CImg and Pandore (in C++):
http://cimg.eu
https://clouard.users.greyc.fr/Pandore/
The CImg Library
CImg is a small open-source C++ library for image processing:
Relatively easy to install and to use.
Generic enough to manage a high variety of image data types
(template-based).
Provide the classical and essential algorithms for image processing.
Extensible library by nature.
Portable on several OS and architectures.
Distributed under an open-source license (CeCILL-C).
Ease scientific reproducibility.
Motivations
Practically, the library is only used by a few hundred experimented programmers.
⇒ Cause: High diversity of people in the image processing field !
Mathematicians
Physicists
Programmers
Biologists ...
⇒ We clearly need more simpler interfaces (than C++ libraries) if we want to
enlarge our audience.
Goals of the G’MIC project
http://gmic.eu
Objectives : G’MIC defines different user interfaces to benefit from CImg features.
Those interfaces are more or less friendly (and powerful) depending on the
user’s skill.
Technical mean : Definition of a comprehensive and concise script language
for the processing of generic image data (G’MIC language). Used as the base
layer for all user interfaces.
slides_gmic.pdf (PDF, 37.2 MB)
Use the permanent link to the download page to share your document on Facebook, Twitter, LinkedIn, or directly with a contact by e-Mail, Messenger, Whatsapp, Line..
Use the short link to share your document on Twitter or by text message (SMS)
Copy the following HTML code to share your document on a Website or Blog