Classify pixels using hyper-spheres or hyper-cubes.

class_ndim_norm classifies the pixels of the multi-channel image given in MultiChannelImage. The result is returned in Regions as one region per classification object. The metric used ('euclid' or 'maximum') is determined by Metric. This paramter must be set to the same value used in learn_ndim_norm. The parameter is used to determine whether one region ('single') or multiple regions ('multiple') have to be generated for each cluster. Radius determines the radii or half edge length of the clusters, respectively. Center determines their centers.

#include  "HalconCpp.h"
#include  <iostream.h>

int main ()
{
  HImage   image ("meer"),
           t1, t2, t3,
           m1, m2, m3, m;

  HWindow  w;

  w.SetColor ("green");
  image.Display (w);

  cout << "Draw your region of interest " << endl;

  HRegion testreg = w.DrawRegion ();

  t1 = image.TextureLaws ("el", 2, 5);     m1 = t1.MeanImage (21, 21);
  t2 = image.TextureLaws ("es", 2, 5);     m2 = t2.MeanImage (21, 21);
  t3 = image.TextureLaws ("le", 2, 5);     m3 = t3.MeanImage (21, 21);

  m  = m1.Compose3 (m2, m3);

  Tuple Metric = "euclid";
  Tuple Radius = 20.0;
  Tuple MinNum = 5;
  Tuple NbrCha = 3;

  HRegion empty;
  Tuple cen, t;

  Radius = testreg.LearnNdimNorm (empty, m, Metric, Radius,
                                  MinNum, NbrCha, &cen, &t);
  Tuple RegMod = "multiple";

  HRegionArray reg = m.ClassNdimNorm (Metric, RegMod, Radius, cen, NbrCha);

  w.SetColored (12);
  reg.Display (w);
  cout << "Result of classification" << endl;
  return (0);
}

Let N be the number of clusters and F be the area of the input region. Then the runtime complexity is O(N * F).

class_ndim_norm returns 2 (H_MSG_TRUE) if all parameters are correct. The behavior with respect to the input images and output regions can be determined by setting the values of the flags 'no_object_result', 'empty_region_result', and 'store_empty_region' with set_system. If necessary, an exception is raised.

class_ndim_norm is reentrant and automatically parallelized (on tuple level).

learn_ndim_norm, compose2, compose3, compose4

connection, select_shape, reduce_domain, select_gray

class_ndim_box, class_2dim_sup

disp_circle, disp_rectangle1

Region processing