main_dep.cu
13 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
#include <stdlib.h>
#include <string>
#include <fstream>
#include <algorithm>
//OpenGL includes
#include <GL/glut.h>
#include <GL/freeglut.h>
//STIM includes
#include <stim/visualization/gl_network.h>
#include <stim/biomodels/network.h>
#include <stim/visualization/gl_aaboundingbox.h>
#include <stim/parser/arguments.h>
#include <stim/visualization/camera.h>
#ifdef __CUDACC__
//CUDA includes
#include <cuda.h>
#endif
//ANN includes
//#include <ANN/ANN.h>
//BOOST includes
#include <boost/tuple/tuple.hpp>
//visualization objects
stim::gl_aaboundingbox<float> bb; //axis-aligned bounding box object
stim::camera cam; //camera object
unsigned num_nets = 0;
stim::gl_network<float> GT; //ground truth network
stim::gl_network<float> T; //test network
//hard-coded parameters
float resample_rate = 0.5f; //sample rate for the network (fraction of sigma used as the maximum sample rate)
float camera_factor = 1.2f; //start point of the camera as a function of X and Y size
float orbit_factor = 0.01f; //degrees per pixel used to orbit the camera
//mouse position tracking
int mouse_x;
int mouse_y;
//OpenGL objects
GLuint cmap_tex = 0; //texture name for the color map
//sets an OpenGL viewport taking up the entire window
void glut_render_single_projection(){
glMatrixMode(GL_PROJECTION); //load the projection matrix for editing
glLoadIdentity(); //start with the identity matrix
int X = glutGet(GLUT_WINDOW_WIDTH); //use the whole screen for rendering
int Y = glutGet(GLUT_WINDOW_HEIGHT);
glViewport(0, 0, X, Y); //specify a viewport for the entire window
float aspect = (float)X / (float)Y; //calculate the aspect ratio
gluPerspective(60, aspect, 0.1, 1000000); //set up a perspective projection
}
//sets an OpenGL viewport taking up the left half of the window
void glut_render_left_projection(){
glMatrixMode(GL_PROJECTION); //load the projection matrix for editing
glLoadIdentity(); //start with the identity matrix
int X = glutGet(GLUT_WINDOW_WIDTH) / 2; //only use half of the screen for the viewport
int Y = glutGet(GLUT_WINDOW_HEIGHT);
glViewport(0, 0, X, Y); //specify the viewport on the left
float aspect = (float)X / (float)Y; //calculate the aspect ratio
gluPerspective(60, aspect, 0.1, 1000000); //set up a perspective projection
}
//sets an OpenGL viewport taking up the right half of the window
void glut_render_right_projection(){
glMatrixMode(GL_PROJECTION); //load the projection matrix for editing
glLoadIdentity(); //start with the identity matrix
int X = glutGet(GLUT_WINDOW_WIDTH) / 2; //only use half of the screen for the viewport
int Y = glutGet(GLUT_WINDOW_HEIGHT);
glViewport(X, 0, X, Y); //specify the viewport on the right
float aspect = (float)X / (float)Y; //calculate the aspect ratio
gluPerspective(60, aspect, 0.1, 1000000); //set up a perspective projection
}
void glut_render_modelview(){
glMatrixMode(GL_MODELVIEW); //load the modelview matrix for editing
glLoadIdentity(); //start with the identity matrix
stim::vec3<float> eye = cam.getPosition(); //get the camera position (eye point)
stim::vec3<float> focus = cam.getLookAt(); //get the camera focal point
stim::vec3<float> up = cam.getUp(); //get the camera "up" orientation
gluLookAt(eye[0], eye[1], eye[2], focus[0], focus[1], focus[2], up[0], up[1], up[2]); //set up the OpenGL camera
}
//draws the network(s)
void glut_render(void) {
if(num_nets == 1){ //if a single network is loaded
glut_render_single_projection(); //fill the entire viewport
glut_render_modelview(); //set up the modelview matrix with camera details
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); //clear the screen
GT.glCenterline(GT.nmags() - 1); //render the GT network (the only one loaded)
}
if(num_nets == 2){ //if two networks are loaded
glut_render_left_projection(); //set up a projection for the left half of the window
glut_render_modelview(); //set up the modelview matrix using camera details
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); //clear the screen
glEnable(GL_TEXTURE_1D); //enable texture mapping
glTexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE); //texture map will be used as the network color
glBindTexture(GL_TEXTURE_1D, cmap_tex); //bind the Brewer texture map
GT.glCenterline(GT.nmags() - 1); //render the GT network
glut_render_right_projection(); //set up a projection for the right half of the window
glut_render_modelview(); //set up the modelview matrix using camera details
T.glCenterline(T.nmags() - 1); //render the T network
}
glutSwapBuffers();
}
// defines camera motion based on mouse dragging
void glut_motion(int x, int y){
float theta = orbit_factor * (mouse_x - x); //determine the number of degrees along the x-axis to rotate
float phi = orbit_factor * (y - mouse_y); //number of degrees along the y-axis to rotate
cam.OrbitFocus(theta, phi); //rotate the camera around the focal point
mouse_x = x; //update the mouse position
mouse_y = y;
glutPostRedisplay(); //re-draw the visualization
}
// sets the mouse position when clicked
void glut_mouse(int button, int state, int x, int y){
mouse_x = x;
mouse_y = y;
}
#define BREWER_CTRL_PTS 11 //number of control points in the Brewer map
void texture_initialize(){
//define the colormap
static float brewer_map[BREWER_CTRL_PTS][3] = { //generate a Brewer color map (blue to red)
{0.192157f, 0.211765f, 0.584314f},
{0.270588f, 0.458824f, 0.705882f},
{0.454902f, 0.678431f, 0.819608f},
{0.670588f, 0.85098f, 0.913725f},
{0.878431f, 0.952941f, 0.972549f},
{1.0f, 1.0f, 0.74902f},
{0.996078f, 0.878431f, 0.564706f},
{0.992157f, 0.682353f, 0.380392f},
{0.956863f, 0.427451f, 0.262745f},
{0.843137f, 0.188235f, 0.152941f},
{0.647059f, 0.0f, 0.14902f}
};
glGenTextures(1, &cmap_tex); //generate a texture map name
glBindTexture(GL_TEXTURE_1D, cmap_tex); //bind the texture map
glTexParameteri(GL_TEXTURE_1D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); //enable linear interpolation
glTexParameteri(GL_TEXTURE_1D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_1D, GL_TEXTURE_WRAP_S, GL_CLAMP); //clamp the values at the minimum and maximum
glTexImage1D(GL_TEXTURE_1D, 0, 3, BREWER_CTRL_PTS, 0, GL_RGB, GL_FLOAT, //upload the texture map to the GPU
brewer_map);
}
//Initialize the OpenGL (GLUT) window, including starting resolution, callbacks, texture maps, and camera
void glut_initialize(){
int myargc = 1; //GLUT requires arguments, so create some bogus ones
char* myargv[1];
myargv [0]=strdup ("netmets");
glutInit(&myargc, myargv); //pass bogus arguments to glutInit()
glutInitDisplayMode(GLUT_DEPTH | GLUT_DOUBLE | GLUT_RGBA); //generate a color buffer, depth buffer, and enable double buffering
glutInitWindowPosition(100,100); //set the initial window position
glutInitWindowSize(320,320); //set the initial window size
glutCreateWindow("NetMets - STIM Lab, UH"); //set the dialog box title
// register callback functions
glutDisplayFunc(glut_render); //function executed for rendering - renders networks
glutMouseFunc(glut_mouse); //executed on a mouse click - sets starting mouse positions for rotations
glutMotionFunc(glut_motion); //executed when the mouse is moved while a button is pressed
texture_initialize(); //set up texture mapping (create texture maps, enable features)
stim::vec3<float> c = bb.center(); //get the center of the network bounding box
//place the camera along the z-axis at a distance determined by the network size along x and y
cam.setPosition(c + stim::vec<float>(0, 0, camera_factor * std::max(bb.size()[0], bb.size()[1])));
cam.LookAt(c[0], c[1], c[2]); //look at the center of the network
glClearColor(1, 1, 1, 1);
}
#ifdef __CUDACC__
void setdevice(int &device){
int count;
cudaGetDeviceCount(&count); // numbers of device that are available
if(count < device + 1){
std::cout<<"No such device available, please set another device"<<std::endl;
exit(1);
}
}
#else
void setdevice(int &device){
device = -1;
}
#endif
//compare both networks and fill the networks with error information
void compare(float sigma, int device){
GT = GT.compare(T, sigma, device); //compare the ground truth to the test case - store errors in GT
T = T.compare(GT, sigma, device); //compare the test case to the ground truth - store errors in T
//calculate the metrics
float FPR = GT.average(0); //calculate the metrics
float FNR = T.average(0);
std::cout << "FNR: " << FPR << std::endl; //print false alarms and misses
std::cout << "FPR: " << FNR << std::endl;
}
// writes features of the networks i.e average segment length, tortuosity, branching index, contraction, fractal dimension, number of end and branch points to a csv file
// Pranathi wrote this - saves network features to a CSV file
void features(std::string filename){
double avgL_t, avgL_gt, avgT_t, avgT_gt, avgB_t, avgB_gt, avgC_t, avgC_gt, avgFD_t, avgFD_gt;
unsigned int e_t, e_gt, b_gt, b_t;
avgL_gt = GT.Lengths();
avgT_gt = GT.Tortuosities();
avgL_t = T.Lengths();
avgT_t = T.Tortuosities();
avgB_gt = GT.BranchingIndex();
avgB_t = T.BranchingIndex();
avgC_gt = GT.Contractions();
avgFD_gt = GT.FractalDimensions();
avgC_t = T.Contractions();
avgFD_t = T.FractalDimensions();
e_gt = GT.EndP();
e_t = T.EndP();
b_gt = GT.BranchP();
b_t = T.BranchP();
std::ofstream myfile;
myfile.open (filename.c_str());
myfile << "Length, Tortuosity, Contraction, Fractal Dimension, Branch Points, End points, Branching Index, \n";
myfile << avgL_gt << "," << avgT_gt << "," << avgC_gt << "," << avgFD_gt << "," << b_gt << "," << e_gt << "," << avgB_gt <<std::endl;
myfile << avgL_t << "," << avgT_t << "," << avgC_t << "," << avgFD_t << "," << b_t << "," << e_t << "," << avgB_t <<std::endl;
myfile.close();
}
// Output an advertisement for the lab, authors, and usage information
void advertise(){
std::cout<<std::endl<<std::endl;
std::cout<<"========================================================================="<<std::endl;
std::cout<<"Thank you for using the NetMets network comparison tool!"<<std::endl;
std::cout<<"Scalable Tissue Imaging and Modeling (STIM) Lab, University of Houston"<<std::endl;
std::cout<<"Developers: Pranathi Vemuri, David Mayerich"<<std::endl;
std::cout<<"Source: https://git.stim.ee.uh.edu/segmentation/netmets"<<std::endl;
std::cout<<"========================================================================="<<std::endl<<std::endl;
std::cout<<"usage: netmets file1 file2 --sigma 3"<<std::endl;
std::cout<<" compare two files with a tolerance of 3 (units defined by the network)"<<std::endl<<std::endl;
std::cout<<" netmets file1 --gui"<<std::endl;
std::cout<<" load a file and display it using OpenGL"<<std::endl<<std::endl;
std::cout<<" netmets file1 file2 --device 0"<<std::endl;
std::cout<<" compare two files using device 0 (if there isn't a gpu, use cpu)"<<std::endl<<std::endl;
}
int main(int argc, char* argv[])
{
stim::arglist args; //create an instance of arglist
//add arguments
args.add("help", "prints this help");
args.add("sigma", "force a sigma value to specify the tolerance of the network comparison", "3");
args.add("gui", "display the network or network comparison using OpenGL");
args.add("device", "choose specific device to run", "0");
args.add("features", "save features to a CSV file, specify file name");
args.parse(argc, argv); //parse the user arguments
if(args["help"].is_set() || args.nargs() == 0){ //test for help
advertise(); //output the advertisement
std::cout<<args.str(); //output arguments
exit(1); //exit
}
if(args.nargs() >= 1){ //if at least one network file is specified
num_nets = 1; //set the number of networks to one
GT.load_obj(args.arg(0)); //load the specified file as the ground truth
/*GT.to_txt("Graph.txt");*/
}
if(args.nargs() == 2){ //if two files are specified, they will be displayed in neighboring viewports and compared
int device = args["device"].as_int(); //get the device value from the user
num_nets = 2; //set the number of networks to two
float sigma = args["sigma"].as_float(); //get the sigma value from the user
T.load_obj(args.arg(1)); //load the second (test) network
if(args["features"].is_set()) //if the user wants to save features
features(args["features"].as_string());
GT = GT.resample(resample_rate * sigma); //resample both networks based on the sigma value
T = T.resample(resample_rate * sigma);
setdevice(device);
compare(sigma, device); //run the comparison algorithm
}
//if a GUI is requested, display the network using OpenGL
if(args["gui"].is_set()){
bb = GT.boundingbox(); //generate a bounding volume
glut_initialize(); //create the GLUT window and set callback functions
glutMainLoop(); // enter GLUT event processing cycle
}
}