int main(int argc,char **argv)
{
cimg_usage("Perona--Malik anisotropic diffusion");
const char* file_i = cimg_option("-i", "input.bmp","Input image");
const char* file_o = cimg_option("-o", (char*)0,"Output image");
int show = cimg_option("-show", 1, "Show progres (0-no, 1-yes");
const double dt = cimg_option("-dt", 0.25, "Time step (if negative, automatic selection)");
const int itm = cimg_option("-it", 100, "Number of Iterations");
const double dta = cimg_option("-delta", 1, "Discontinuity parameter");
CImg<float> image_in(file_i);
CImg<float> image = image_in.resize(-100,-100,-100, 1); // Convert to grayscale image
CImg<float> result(image);
result.fill(0);
CImgDisplay* disp_input=NULL;
if (show==1) disp_input = new CImgDisplay(image, "Input image");
result = get_perona_malik2d(image, itm, dt, dta, show);
if (disp_input) {
while (!disp_input->is_closed()) disp_input->wait();
}
if (file_o) result.save(file_o);
return 0;
}
// Main procedure
//---------------
int main(int argc,char **argv) {
// Read and check command line parameters.
cimg_usage("Compute a linear fading between two 2D images");
const char *file_i1 = cimg_option("-i1",cimg_imagepath "sh0r.pgm","Input Image 1");
const char *file_i2 = cimg_option("-i2",cimg_imagepath "milla.bmp","Input Image 2");
const char *file_o = cimg_option("-o",(char*)0,"Output Image");
const bool visu = cimg_option("-visu",true,"Visualization mode");
const double pmin = cimg_option("-min",40.0,"Begin of the fade (in %)")/100.0;
const double pmax = cimg_option("-max",60.0,"End of the fade (in %)")/100.0;
const double angle = cimg_option("-angle",0.0,"Fade angle")*cil::cimg::PI/180;
// Init images.
cil::CImg<unsigned char> img1(file_i1), img2(file_i2);
if (!img2.is_sameXYZC(img1)) {
int
dx = std::max(img1.width(),img2.width()),
dy = std::max(img1.height(),img2.height()),
dz = std::max(img1.depth(),img2.depth()),
dv = std::max(img1.spectrum(),img2.spectrum());
img1.resize(dx,dy,dz,dv,3);
img2.resize(dx,dy,dz,dv,3);
}
cil::CImg<unsigned char> dest(img1);
// Compute the faded image.
const double ca = std::cos(angle), sa = std::sin(angle);
double alpha;
cimg_forXYZC(dest,x,y,z,k) {
const double X = ((double)x/img1.width() - 0.5)*ca + ((double)y/img1.height() - 0.5)*sa;
if (X + 0.5<pmin) alpha = 0; else {
if (X + 0.5>pmax) alpha = 1; else
alpha = (X + 0.5 - pmin)/(pmax - pmin);
}
dest(x,y,z,k) = (unsigned char)((1 - alpha)*img1(x,y,z,k) + alpha*img2(x,y,z,k));
}
// Save and exit
if (file_o) dest.save(file_o);
if (visu) dest.display("Image fading");
return 0;
}
int main(int argc,char **argv)
{
cimg_usage("Search Path");
const char* file_i = cimg_option("-i","input.bmp","Input image");
const int blur = cimg_option("-blur",2,"CImg blur value");
// float sigma = cimg_option("-s", 1.0,"Sigma");
CImg<unsigned char> image_in(file_i);
CImg<unsigned char> image_gray = image_in.get_resize(-100,-100,-100,1);
CImg<unsigned char> image_rgb = image_in.get_resize(-100,-100,-100,3);
CImg<unsigned char> blur_img = image_gray.get_blur(blur);
// CImg<unsigned char> blur_img(image);
// blur_img.fill(0);
// PGaussianFiltering2D(image, blur_img, sigma);
CImg<unsigned char> gradient = gradient_cimg(blur_img, 2);
CImg<unsigned char> gradient_norm = gradient.get_normalize(10,255);
CImg<unsigned char> gradient_colr = gradient_cimg( image_rgb.get_blur(blur) , 2);
int Dx = image_in.width();
int Dy = image_in.height();
CImg<unsigned char> canvas(image_rgb);
CImg<unsigned char> gcanvas(gradient_colr);
CImg<unsigned char> marker(Dx,Dy,1,1,0);
CImg<unsigned char> marker_neg(Dx,Dy,1,1,0);
CImg<unsigned char> gradmarker(Dx,Dy,1,1,0);
CImg<signed int> result(Dx,Dy,1,1,-1);
CImg<unsigned char> result_ok0(Dx,Dy,1,1,0);
CImg<unsigned char> result_ok1(Dx,Dy,1,3,0);
CImg<unsigned char> result_ok2(Dx,Dy,1,1,0);
CImgDisplay disp_canvas( cimg_fitscreen(canvas.width(),canvas.height(),canvas.depth()), "Usage: L,R mouse, F1,F2,c" );
CImgDisplay disp_result( cimg_fitscreen(canvas.width(),canvas.height(),canvas.depth()), "Result" );
CImgDisplay disp_gcanvas( cimg_fitscreen(canvas.width(),canvas.height(),canvas.depth()), "Gradient" );
canvas.display(disp_canvas);
gcanvas.display(disp_gcanvas);
result_ok0.display(disp_result);
CImg<unsigned char> marker_disp(disp_canvas.width(), disp_canvas.height(), 1, 3, 0);
int circle_size1a = 3;
int circle_size1b = (float) Dx / (float)disp_canvas.width() * circle_size1a + 0.5f;
int circle_size2a = 5;
int circle_size2b = (float) Dx / (float)disp_canvas.width() * circle_size2a + 0.5f;
randomColorForLabels();
int draw_type=1;
bool clicked=false;
volatile unsigned int btt=0;
int mx=0, my=0, px=0, py=0, mouse_x=0, mouse_y=0, prev_x=0, prev_y=0;
while (!disp_canvas.is_closed()) {
disp_canvas.wait();
if (disp_canvas.key()) {
switch (disp_canvas.key()) {
case cimg::keyF1: draw_type=1; break;
case cimg::keyF2: draw_type=2; break;
case cimg::keyC:
marker.fill(0); marker_disp.fill(0);
canvas.assign(image_rgb).display(disp_canvas);
gcanvas.assign(gradient_colr).display(disp_gcanvas);
break;
}
disp_canvas.set_key();
}
if (disp_canvas.mouse_x()>=0 && disp_canvas.mouse_y()>=0) {
mx = disp_canvas.mouse_x(),
my = disp_canvas.mouse_y();
mouse_x = mx*Dx/disp_canvas.width();
mouse_y = my*Dy/disp_canvas.height();
btt = disp_canvas.button();
if (btt == 1) {
if (draw_type==1) {
marker.draw_line(mouse_x,mouse_y,prev_x,prev_y,white);
marker_disp.draw_line(mx,my,px,py,red);
}
else {
marker.draw_circle(mouse_x,mouse_y,circle_size1b,white);
marker_disp.draw_circle(mx,my,circle_size1a,red);
}
draw_results(canvas,marker,red); //canvas = canvas | marker;
draw_results(gcanvas,marker,red); //gcanvas = gcanvas | marker;
clicked=true;
} else
if (btt == 2) {
marker.draw_circle(mouse_x,mouse_y,circle_size2b,black);
marker_disp.draw_circle(mx,my,circle_size2a,white);
canvas.assign(image_rgb); //.resize(disp_canvas);
draw_results(canvas,marker,red);
draw_results(canvas,result_ok2,yellow);
gcanvas.assign(gradient_colr); //.resize(disp_gcanvas);
draw_results(gcanvas,marker,red);
draw_results(gcanvas,result_ok2,yellow);
clicked=true;
} else
if (clicked) {
cimg_forXY(marker,x,y) marker_neg(x,y) = 255-marker(x,y); // Invert marker images (lines should to be black)
gradmarker = gradient_norm & marker_neg; // Add marker image to gradient image
result.fill(-1);
WatershedSegmentation(gradmarker, &result, 1, 0); // Run marker-based watershed algorithm
//result_ok0 = result.equalize(256,0,255);
result_ok1 = colorizeLabelImage( result );
result_ok2 = gradient_cimg(result,2); // Boundary detection
cimg_forXY(result_ok2,x,y) result_ok2(x,y) = result_ok2(x,y)>0?255:0;
canvas.assign(image_rgb); //.resize(disp_canvas);
draw_results(canvas,marker,red);
draw_results(canvas,result_ok2,yellow);
gcanvas.assign(gradient_colr); //.resize(disp_gcanvas);
draw_results(gcanvas,marker,red);
draw_results(gcanvas,result_ok2,yellow);
clicked=false;
}
prev_x = mouse_x; prev_y = mouse_y;
px = mx; py = my;
canvas.display(disp_canvas);
gcanvas.display(disp_gcanvas);
result_ok1.display(disp_result);
}
}
return 0;
}
