Difference between revisions of "Publications/crozet.14.icip.inc"

From LRDE
 
(6 intermediate revisions by the same user not shown)
Line 1: Line 1:
  +
__TOC__
  +
 
== Figures ==
  
== Figures ==
  +
  ⚫
=== Fig. 1 ===
  +
  +
Sample uses of the tree of shapes.
   
 
{| class="wikitable" border="1"
  
{| class="wikitable" border="1"
 
|-
  
|-
−
| [[File:Crozet14icip_Intronoise.png|150px|x]]
 +
| [[File:Crozet14icip_Intronoise.png|100px|x]]
−
| [[File:Crozet14icip_Intronoiseout.png|150px|x‎]]
 +
| [[File:Crozet14icip_Intronoiseout.png|100px|x‎]]
 
|-
  
|-
 
! colspan="2" | (a) Denoising (self-dual grain removal).
  
! colspan="2" | (a) Denoising (self-dual grain removal).
 
|-
  
|-
−
| [[File:Crozet14icip_Introzeolite.png|150px|x]]
 +
| [[File:Crozet14icip_Introzeolite.png|x]]
−
| [[File:Crozet14icip_Introzeoliteout.png‎|150px|x]]
 +
| [[File:Crozet14icip_Introzeoliteout.png‎|x]]
 
|-
  
|-
 
! colspan="2" | (b) Shape Filtering (keep round objects).
  
! colspan="2" | (b) Shape Filtering (keep round objects).
 
|-
  
|-
−
| [[File:Crozet14icip_Introsynthetic.png|300px|x]]
 +
| [[File:Crozet14icip_Introsynthetic.png|none|x]]
−
| [[File:Crozet14icip_Introsyntheticout.png‎|300px|x]]
 +
| [[File:Crozet14icip_Introsyntheticout.png‎|none|x]]
 
|-
  
|-
 
! colspan="2" | (c) Object Detection (energy-based method).
  
! colspan="2" | (c) Object Detection (energy-based method).
 
|-
  
|-
−
| [[File:Crozet14icip_Introplane.png|300px|x]]
 +
| [[File:Crozet14icip_Introplane.png|none|x]]
−
| [[File:Crozet14icip_Introplanehierarchy.png|thumb|300px‎|x]]
 +
| [[File:Crozet14icip_Introplanehierarchy.png|thumb|x]]
 
|-
  
|-
−
! colspan="2" | (d) Hierarchical Segmentation (saliency-based).
 +
! colspan="2" | (d) Hierarchical Segmentation (saliency-based): ''click on the thumbnail (right)''.
 
|-
  
|-
−
| [[File:Crozet14icip_Introplanesegmentationfine.png|300px|x]]
 +
| [[File:Crozet14icip_Introplanesegmentationfine.png|none|x]]
−
| [[File:Crozet14icip_Introplanesegmentationcoarse.png|thumb|300px‎|x]]
 +
| [[File:Crozet14icip_Introplanesegmentationcoarse.png|none|x]]
 
|-
  
|-
−
! colspan="2" | (d') Hierarchical Segmentation (saliency-based).
 +
! colspan="2" | (e) Hierarchical Segmentation: fine (left), coarse (right).
 
|}
  
|}
−
'''Fig. 1: Sample uses of the tree of shapes (left column: input images; right column: state-of-the-art results).'''
  
   
  +
  ⚫
=== Fig. 2 ===
  +
  +
An image (a) and its tree of shapes (b). The propagation of the level line λ ended, meaning that the nodes O and A have already been visited. The hierarchical queue contains the interior contour of B and C. Thus it can be partitioned in two sets S⁺λ = ∂B and S⁻λ = ∂C. The propagation can proceed on both parts in parallel.
   
 
{| class="wikitable" border="1"
  
{| class="wikitable" border="1"
Line 39: Line 48:
 
! (b) Tree of shapes
  
! (b) Tree of shapes
 
|}
  
|}
⚫
'''Fig. 2'''
  
   
  +
=== Fig. 4 ===
  +
  +
(a) is the input image. (b) is the result of the subdivision. (c) is the result of the immersion into the Khalimsky grid. 0-faces are represented by dots, 1-faces by segments and 2-faces by squares.
  +
  +
{| class="wikitable" border="1"
  +
|-
  +
| [[File:Crozet14icip_Immerse_f.png|90px]]
  +
| ~ [[File:Crozet14icip_Immerse_f_subdivided‎.png|160px]] ~
  +
| ~ [[File:Crozet14icip_Immerse_f_immersed‎.png|160px]] ~
  +
|-
  +
! (a) Input
  +
! (b) Subdivided
  +
! (c) Immersed
  +
|}
  +
  +
  +
=== Fig. 8 ===
  +
  +
The original image (a) and the associated F^{ord} (b); the max-tree of (b) coincides with the tree of shapes of (a).
  +
  +
{| class="wikitable" border="1"
  +
|-
  +
| [[File:Crozet14icip_Simpleimage_levels.png]]
  +
| [[File:Crozet14icip_Simpleimage_revalued.png]]
  +
|-
  +
! (a) Original image
  +
! (b) Re-valued image
  +
|}
  +
  +
  +
=== Fig. 10 ===
  +
  +
Computation times (in seconds) on a classical image test set of the following algorithms: FLLT, FLST, Géraud et al., and this paper proposal.
  +
  +
{| class="wikitable" border="1"
  +
|-
  +
| [[File:Crozet14icip_Benchwo.png]]
  +
|}
   
   
 
== Images ==
  
== Images ==
⚫
Images used for the benchmarks: [https://www.dropbox.com/s/ff3gfhiivalcjyz/images.tar.bz2]
  
   
  ⚫
Images used for the benchmarks: [http://www.lrde.epita.fr/~theo/papers/crozet14icip_materials/images.tar.bz2]
⚫
== Code ==
  
  +
⚫
* Code of the serial version: [https://www.dropbox.com/s/w7rje3w4wq6pi4d/serial.tar.bz2]
  
  +
⚫
* Code of the parallel version: [https://www.dropbox.com/s/ytslb3d73jm5owa/parallel.tar.bz2]
  
  +
== Source Code ==
⚫
* Code of the milena image processing library: [https://www.dropbox.com/s/3dmv3hg0h3561y4/mln.tar.bz2]
  
  +
  ⚫
* Code of the serial version: [http://www.lrde.epita.fr/~theo/papers/crozet14icip_materials/serial.tar.bz2]
  ⚫
* Code of the parallel version: [http://www.lrde.epita.fr/~theo/papers/crozet14icip_materials/parallel.tar.bz2]
  ⚫
* Code of the milena image processing library: [http://www.lrde.epita.fr/~theo/papers/crozet14icip_materials/mln.tar.bz2]
  +
  +
  +
== Useful links ==
  +
  +
The ''Olena'' platform for image processing: http://olena.lrde.epita.fr
  +
(containing the ''Milena'' C++ image processing library)
  +
  +
Reproducible research:
  +
* http://reproducibleresearch.net/index.php/Main_Page
  +
* http://en.wikipedia.org/wiki/Reproducibility#Reproducible_research
  +
  +
  +
== Contact ==
  +
  +
* Homepage: http://www.lrde.epita.fr/wiki/User:Theo
  +
* Email: [mailto:thierry.geraud@lrde.epita.fr thierry.geraud@lrde.epita.fr]

Latest revision as of 15:46, 18 February 2014

Figures

Fig. 1

Sample uses of the tree of shapes.

x x‎
(a) Denoising (self-dual grain removal).
x x
(b) Shape Filtering (keep round objects).
x
x
(c) Object Detection (energy-based method).
x
x
(d) Hierarchical Segmentation (saliency-based): click on the thumbnail (right).
x
x
(e) Hierarchical Segmentation: fine (left), coarse (right).


Fig. 2

An image (a) and its tree of shapes (b). The propagation of the level line λ ended, meaning that the nodes O and A have already been visited. The hierarchical queue contains the interior contour of B and C. Thus it can be partitioned in two sets S⁺λ = ∂B and S⁻λ = ∂C. The propagation can proceed on both parts in parallel.

Crozet14icip simpleimage.png Crozet14icip simpletos.png
(a) Image (b) Tree of shapes

Fig. 4

(a) is the input image. (b) is the result of the subdivision. (c) is the result of the immersion into the Khalimsky grid. 0-faces are represented by dots, 1-faces by segments and 2-faces by squares.

Crozet14icip Immerse f.png ~ Crozet14icip Immerse f subdivided.png ~ ~ Crozet14icip Immerse f immersed.png ~
(a) Input (b) Subdivided (c) Immersed


Fig. 8

The original image (a) and the associated F^{ord} (b); the max-tree of (b) coincides with the tree of shapes of (a).

Crozet14icip Simpleimage levels.png Crozet14icip Simpleimage revalued.png
(a) Original image (b) Re-valued image


Fig. 10

Computation times (in seconds) on a classical image test set of the following algorithms: FLLT, FLST, Géraud et al., and this paper proposal.

Crozet14icip Benchwo.png


Images

Images used for the benchmarks: [1]


Source Code

  • Code of the serial version: [2]
  • Code of the parallel version: [3]
  • Code of the milena image processing library: [4]


Useful links

The Olena platform for image processing: http://olena.lrde.epita.fr (containing the Milena C++ image processing library)

Reproducible research:


Contact

Retrieved from "https://www.lrde.epita.fr/index.php?title=Publications/crozet.14.icip.inc&oldid=6972"