1 /*
2  #
3  #  File        : use_skeleton.cpp
4  #                ( C++ source file )
5  #
6  #  Description : Example of use for the CImg plugin 'plugins/skeleton.h'.
7  #                This file is a part of the CImg Library project.
8  #                ( http://cimg.eu )
9  #
10  #  Copyright   : Francois-Xavier Dupe
11  #                ( http://www.greyc.ensicaen.fr/~fdupe/ )
12  #
13  #  License     : CeCILL v2.0
14  #                ( http://www.cecill.info/licences/Licence_CeCILL_V2-en.html )
15  #
16  #  This software is governed by the CeCILL  license under French law and
17  #  abiding by the rules of distribution of free software.  You can  use,
18  #  modify and/ or redistribute the software under the terms of the CeCILL
19  #  license as circulated by CEA, CNRS and INRIA at the following URL
20  #  "http://www.cecill.info".
21  #
22  #  As a counterpart to the access to the source code and rights to copy,
23  #  modify and redistribute granted by the license, users are provided only
24  #  with a limited warranty  and the software's author,  the holder of the
25  #  economic rights,  and the successive licensors  have only  limited
26  #  liability.
27  #
28  #  In this respect, the user's attention is drawn to the risks associated
29  #  with loading,  using,  modifying and/or developing or reproducing the
30  #  software by the user in light of its specific status of free software,
31  #  that may mean  that it is complicated to manipulate,  and  that  also
32  #  therefore means  that it is reserved for developers  and  experienced
33  #  professionals having in-depth computer knowledge. Users are therefore
34  #  encouraged to load and test the software's suitability as regards their
35  #  requirements in conditions enabling the security of their systems and/or
36  #  data to be ensured and,  more generally, to use and operate it in the
37  #  same conditions as regards security.
38  #
39  #  The fact that you are presently reading this means that you have had
40  #  knowledge of the CeCILL license and that you accept its terms.
41  #
42 */
43 
44 #include <queue>
45 #define cimg_plugin "plugins/skeleton.h"
46 #include "CImg.h"
47 using namespace cimg_library;
48 #ifndef cimg_imagepath
49 #define cimg_imagepath "img/"
50 #endif
51 
52 // Main procedure
53 //----------------
main(int argc,char ** argv)54 int main (int argc, char **argv) {
55 
56   cimg_usage("Compute the skeleton of a shape, using Hamilton-Jacobi equations");
57 
58   // Read command line arguments
59   cimg_help("Input/Output options\n"
60             "--------------------");
61   const char* file_i = cimg_option("-i",cimg_imagepath "milla.bmp","Input (black&white) image");
62   const int median = cimg_option("-median",0,"Apply median filter");
63   const bool invert = cimg_option("-inv",false,"Invert image values");
64   const char* file_o = cimg_option("-o",(char*)0,"Output skeleton image");
65   const bool display = cimg_option("-visu",true,"Display results");
66 
67   cimg_help("Skeleton computation parameters\n"
68             "-------------------------------");
69   const float thresh = cimg_option("-t",-0.3f,"Threshold");
70   const bool curve = cimg_option("-curve",false,"Create medial curve");
71 
72   cimg_help("Torsello correction parameters\n"
73             "------------------------------");
74   const bool correction = cimg_option("-corr",false,"Torsello correction");
75   const float dlt1 = 2;
76   const float dlt2 = cimg_option("-dlt",1.0f,"Discrete step");
77 
78   // Load the image (forcing it to be scalar with 2 values { 0,1 }).
79   CImg<unsigned int> image0(file_i), image = image0.get_norm().quantize(2).normalize(0.0f,1.0f).round();
80   if (median) image.blur_median(median);
81   if (invert) (image-=1)*=-1;
82   if (display) (image0.get_normalize(0,255),image.get_normalize(0,255)).display("Input image - Binary image");
83 
84   // Compute distance map.
85   CImgList<float> visu;
86   CImg<float> distance = image.get_distance(0);
87   if (display) visu.insert(distance);
88 
89   // Compute the gradient of the distance function, and the flux (divergence) of the gradient field.
90   const CImgList<float> grad = distance.get_gradient("xyz");
91   CImg<float> flux = image.get_flux(grad,1,1);
92   if (display) visu.insert(flux);
93 
94   // Use the Torsello correction of the flux if necessary.
95   if (correction) {
96     CImg<float>
97       logdensity = image.get_logdensity(distance,grad,flux,dlt1),
98       nflux = image.get_corrected_flux(logdensity,grad,flux,dlt2);
99     if (display) visu.insert(logdensity).insert(nflux);
100     flux = nflux;
101   }
102 
103   if (visu) {
104     cimglist_apply(visu,normalize)(0,255);
105     visu.display(visu.size()==2?"Distance function - Flux":"Distance function - Flux - Log-density - Corrected flux");
106   }
107 
108   // Compute the skeleton
109   const CImg<unsigned int> skel = image.get_skeleton(flux,distance,curve,thresh);
110   if (display) {
111     (image0.resize(-100,-100,1,3)*=0.7f).get_shared_channel(1)|=skel*255.0;
112     image0.draw_image(0,0,0,0,image*255.0,0.5f).display("Image + Skeleton");
113   }
114 
115   // Save output image if necessary.
116   if (file_o) skel.save(file_o);
117 
118   return 0;
119 }
120