Several modifications in segmentation, EM, k-means:

segmentation : moving center points which falls wrong side of the facet.
EM: random initialization

Surface_mesh_segmentation/include/CGAL/Surface_mesh_segmentation.h

@@ -14,6 +14,20 @@
  * +) Deciding how to generate rays in cone: for now using "polar angle" and "accept-reject (square)" and "concentric mapping" techniques
  */
 
+
+//#include "Expectation_maximization.h"
+//#include "K_means_clustering.h"
+//#include "Timer.h"
+
+#include <CGAL/internal/Surface_mesh_segmentation/Expectation_maximization.h>
+#include <CGAL/internal/Surface_mesh_segmentation/K_means_clustering.h>
+
+#include <CGAL/Simple_cartesian.h>
+#include <CGAL/AABB_tree.h>
+#include <CGAL/AABB_traits.h>
+#include <CGAL/AABB_polyhedron_triangle_primitive.h>
+#include <CGAL/utility.h>
+
 #include <iostream>
 #include <fstream>
 #include <cmath>
@@ -21,22 +35,17 @@
 #include <algorithm>
 #include <utility>
 
-//#include "Timer.h"
-//#include "Expectation_maximization.h"
-//#include "K_means_clustering.h"
-#include <CGAL/Simple_cartesian.h>
-#include <CGAL/internal/Surface_mesh_segmentation/Expectation_maximization.h>
-#include <CGAL/internal/Surface_mesh_segmentation/K_means_clustering.h>
-
-#include <CGAL/AABB_tree.h>
-#include <CGAL/AABB_traits.h>
-#include <CGAL/AABB_polyhedron_triangle_primitive.h>
-#include <CGAL/utility.h>
-
 #define LOG_5 1.60943791
 #define NORMALIZATION_ALPHA 4.0
 #define ANGLE_ST_DEV_DIVIDER 3.0
 
+//#define ACTIVATE_SEGMENTATION_DEBUG
+#ifdef ACTIVATE_SEGMENTATION_DEBUG
+#define SEG_DEBUG(x) x;
+#else
+#define SEG_DEBUG(x)
+#endif
+
 namespace CGAL
 {
 
@@ -53,6 +62,8 @@ public:
   typedef typename Polyhedron::Facet_handle   Facet_handle;
 protected:
   typedef typename Kernel::Ray_3   Ray;
+  typedef typename Kernel::Plane_3 Plane;
+
   typedef typename CGAL::AABB_polyhedron_triangle_primitive<Kernel, Polyhedron>
   Primitive;
   typedef typename CGAL::AABB_tree<CGAL::AABB_traits<Kernel, Primitive> >
@@ -105,8 +116,7 @@ public:
   void calculate_sdf_values();
 
   double calculate_sdf_value_of_facet (const Facet_handle& facet,
-                                       const Point& center,
+                                       const Tree& tree) const;
-                                       const Vector& normal_const, const Tree& tree) const;
   void cast_and_return_minimum (const Ray& ray, const Tree& tree,
                                 const Facet_handle& facet,
                                 bool& is_found, double& min_distance) const;
@@ -118,6 +128,9 @@ public:
   double calculate_sdf_value_from_rays_with_trimmed_mean (
     std::vector<double>& ray_distances, std::vector<double>& ray_weights) const;
 
+  void arrange_center_orientation(const Plane& plane, const Vector& unit_normal,
+                                  Point& center) const;
+
   void disk_sampling_rejection();
   void disk_sampling_polar_mapping();
   void disk_sampling_concentric_mapping();
@@ -141,13 +154,14 @@ inline Surface_mesh_segmentation<Polyhedron>::Surface_mesh_segmentation(
   : mesh(mesh), cone_angle(cone_angle), number_of_rays_sqrt(number_of_rays_sqrt),
     number_of_centers(number_of_centers), log_file("log_file.txt")
 {
-  //Timer t;
+  SEG_DEBUG(Timer t)
+
   disk_sampling_concentric_mapping();
   calculate_sdf_values();
-  //std::cout << t;
+  SEG_DEBUG(std::cout << t)
-  apply_GMM_fitting_with_K_means_init();
+  apply_GMM_fitting();
   //write_sdf_values("sdf_values_sample_dino_2.txt");
-  //read_sdf_values("sdf_values_sample_camel.txt");
+  //read_sdf_values("sdf_values_sample_dino.txt");
 }
 
 template <class Polyhedron>
@@ -159,14 +173,8 @@ inline void Surface_mesh_segmentation<Polyhedron>::calculate_sdf_values()
       facet_it != mesh->facets_end(); ++facet_it) {
     CGAL_precondition(facet_it->is_triangle()); //Mesh should contain triangles.
 
-    Point v1 = facet_it->halfedge()->vertex()->point();
-    Point v2 = facet_it->halfedge()->next()->vertex()->point();
-    Point v3 = facet_it->halfedge()->next()->next()->vertex()->point();
-    Point center  = CGAL::centroid(v1, v2, v3);
-    Vector normal = CGAL::unit_normal(v1, v2,
-                                      v3) * -1.0; //Assuming triangles are CCW oriented.
     //SL: cone angle and number of rays should be parameters.
-    double sdf = calculate_sdf_value_of_facet(facet_it, center, normal, tree);
+    double sdf = calculate_sdf_value_of_facet(facet_it, tree);
     sdf_values.insert(std::pair<Facet_handle, double>(facet_it, sdf));
   }
   normalize_sdf_values();
@@ -176,14 +184,22 @@ inline void Surface_mesh_segmentation<Polyhedron>::calculate_sdf_values()
 template <class Polyhedron>
 inline double
 Surface_mesh_segmentation<Polyhedron>::calculate_sdf_value_of_facet(
-  const Facet_handle& facet, const Point& center,
+  const Facet_handle& facet, const Tree& tree) const
-  const Vector& normal_const, const Tree& tree) const
 {
-  typename Kernel::Plane_3 plane(center, normal_const);
+  Point p1 = facet->halfedge()->vertex()->point();
+  Point p2 = facet->halfedge()->next()->vertex()->point();
+  Point p3 = facet->halfedge()->next()->next()->vertex()->point();
+  Point center  = CGAL::centroid(p1, p2, p3);
+  Vector normal = CGAL::unit_normal(p1, p2,
+                                    p3) * -1.0; //Assuming triangles are CCW oriented.
+
+  Plane plane(center, normal);
   Vector v1 = plane.base1();
   Vector v2 = plane.base2();
-  v1 = v1 / CGAL::sqrt(v1.squared_length());
+  v1 = v1 / sqrt(v1.squared_length());
-  v2 = v2 / CGAL::sqrt(v2.squared_length());
+  v2 = v2 / sqrt(v2.squared_length());
+
+  arrange_center_orientation(plane, normal, center);
 
   int ray_count = number_of_rays_sqrt * number_of_rays_sqrt;
 
@@ -192,7 +208,7 @@ Surface_mesh_segmentation<Polyhedron>::calculate_sdf_value_of_facet(
   ray_weights.reserve(ray_count);
 
   double length_of_normal = 1.0 / tan(cone_angle / 2);
-  Vector normal = normal_const * length_of_normal;
+  normal = normal * length_of_normal;
 
   for(Disk_samples_list::const_iterator sample_it = disk_samples.begin();
       sample_it != disk_samples.end(); ++sample_it) {
@@ -256,6 +272,8 @@ void Surface_mesh_segmentation<Polyhedron>::cast_and_return_minimum(
                  CGAL::ORIGIN + min_normal) != CGAL::ACUTE) {
     is_found = false;
   }
+  //total_intersection += intersections.size();
+  //total_cast++;
 }
 
 template <class Polyhedron>
@@ -403,6 +421,49 @@ Surface_mesh_segmentation<Polyhedron>::calculate_sdf_value_from_rays_with_trimme
   return total_distance / total_weights;
 }
 
+/* Slightly move center towards inverse normal of facet.
+ * Parameter plane is constructed by inverse normal.
+ */
+template <class Polyhedron>
+void Surface_mesh_segmentation<Polyhedron>::arrange_center_orientation(
+  const Plane& plane, const Vector& unit_normal, Point& center) const
+{
+  /*
+  *  Not sure how to decide on how much I should move center ?
+  */
+  double epsilon = 1e-8;
+  // Option-1
+  //if(plane.has_on_positive_side(center)) { return; }
+  //Vector epsilon_normal = unit_normal * epsilon;
+  //do {
+  //    center = CGAL::operator+(center, epsilon_normal);
+  //} while(!plane.has_on_positive_side(center));
+
+  //Option-2
+  //if(plane.has_on_positive_side(center)) { return; }
+  //double distance = sqrt(CGAL::squared_distance(plane, center));
+  //distance = distance > epsilon ? (distance + epsilon) : epsilon;
+  //Vector distance_normal = unit_normal * distance;
+  //
+  //do {
+  //    center = CGAL::operator+(center, distance_normal);
+  //} while(!plane.has_on_positive_side(center));
+
+  //Option-3
+  Ray ray(center, unit_normal);
+  Kernel::Do_intersect_3 intersector = Kernel().do_intersect_3_object();
+  if(!intersector(ray, plane)) {
+    return;
+  }
+
+  Vector epsilon_normal = unit_normal * epsilon;
+  do {
+    center = CGAL::operator+(center, epsilon_normal);
+    ray = Ray(center, unit_normal);
+  } while(intersector(ray, plane));
+
+}
+
 template <class Polyhedron>
 inline void Surface_mesh_segmentation<Polyhedron>::disk_sampling_rejection()
 {
@@ -522,7 +583,7 @@ inline void Surface_mesh_segmentation<Polyhedron>::smooth_sdf_values()
       total_neighbor_sdf += sdf_values[facet_circulator->opposite()->facet()];
     } while( ++facet_circulator !=  f->facet_begin());
     total_neighbor_sdf /= 3;
-    smoothed_sdf_values[f] = (sdf_values[f] + total_neighbor_sdf) / 2;
+    smoothed_sdf_values[f] = (pair_it->second + total_neighbor_sdf) / 2;
   }
   sdf_values = smoothed_sdf_values;
 }
@@ -610,6 +671,7 @@ inline void Surface_mesh_segmentation<Polyhedron>::read_sdf_values(
   const char* file_name)
 {
   std::ifstream input(file_name);
+  sdf_values.clear();
   for(Facet_iterator facet_it = mesh->facets_begin();
       facet_it != mesh->facets_end(); ++facet_it) {
     double sdf_value;
@@ -640,4 +702,8 @@ inline void Surface_mesh_segmentation<Polyhedron>::read_center_ids(
 #undef ANGLE_ST_DEV_DIVIDER
 #undef LOG_5
 #undef NORMALIZATION_ALPHA
+
+#ifdef SEG_DEBUG
+#undef SEG_DEBUG
+#endif
 #endif //CGAL_SURFACE_MESH_SEGMENTATION_H

Surface_mesh_segmentation/include/CGAL/internal/Surface_mesh_segmentation/Expectation_maximization.h

@@ -2,7 +2,7 @@
 #define CGAL_SEGMENTATION_EXPECTATION_MAXIMIZATION_H
 /* NEED TO BE DONE */
 /* About implementation:
- * Calculating probability multiple times
+ * Calculating probability multiple times, solution: storing matrix(cluster, point) for probability values.
  * About safe division: where centers have no members, in graph cut it may happen
  */
 #include <vector>
@@ -12,6 +12,15 @@
 #include <cstdlib>
 #include <limits>
 
+#include <fstream>
+#include <iostream>
+//#define ACTIVATE_SEGMENTATION_EM_DEBUG
+#ifdef ACTIVATE_SEGMENTATION_EM_DEBUG
+#define SEG_DEBUG(x) x;
+#else
+#define SEG_DEBUG(x)
+#endif
+
 namespace CGAL
 {
 
@@ -102,14 +111,17 @@ protected:
 public:
   Expectation_maximization(int number_of_centers, const std::vector<double>& data,
                            const std::vector<int>& initial_centers = std::vector<int>(),
-                           int maximum_iteration = 100)
+                           int number_of_runs = 50, int maximum_iteration = 100)
-    : points(data.begin(), data.end()), threshold(1e-3),
+    : points(data.begin(), data.end()), threshold(1e-4),
       maximum_iteration(maximum_iteration), is_converged(false), seed(time(NULL)) {
     srand(seed);
-    //calculate_fitting_with_multiple_run(number_of_centers, 50);
+    if(initial_centers.empty()) {
+      calculate_fitting_with_multiple_run(number_of_centers, number_of_runs);
+    } else {
       initiate_centers(number_of_centers, initial_centers);
       calculate_fitting();
     }
+  }
 
   void fill_with_center_ids(std::vector<int>& data_centers) {
     data_centers.reserve(points.size());
@@ -133,21 +145,26 @@ protected:
   void initiate_centers_randomly(int number_of_centers) {
     centers.clear();
     /* Randomly generate means of centers */
+    //vector<int> center_indexes;
+    double initial_mixing_coefficient = 1.0;
     double initial_deviation = 1.0  / (2.0 * number_of_centers);
-    double initial_mixing_coefficient = 1.0 / number_of_centers;
     for(int i = 0; i < number_of_centers; ++i) {
-      double initial_mean = points[rand() % points.size()].data;
+      double random_index = rand() % points.size();
+      //center_indexes.push_back(random_index);
+      Gaussian_point mean_point = points[random_index];
+      double initial_mean = mean_point.data;
       centers.push_back(Gaussian_center(initial_mean, initial_deviation,
                                         initial_mixing_coefficient));
     }
     sort(centers.begin(), centers.end());
+    //write_random_centers("center_indexes.txt", center_indexes);
   }
 
   void initiate_centers_uniformly(int number_of_centers) {
     centers.clear();
     /* Uniformly generate centers */
     double initial_deviation = 1.0  / (2.0 * number_of_centers);
-    double initial_mixing_coefficient = 1.0 / number_of_centers;
+    double initial_mixing_coefficient = 1.0;
     for(int i = 0; i < number_of_centers; ++i) {
       double initial_mean = (i + 1.0) / (number_of_centers + 1.0);
       centers.push_back(Gaussian_center(initial_mean, initial_deviation,
@@ -203,6 +220,14 @@ protected:
       point_it->calculate_total_membership(centers);
     }
   }
+  double calculate_deviation(const Gaussian_point& point) const {
+    double deviation = 0.0;
+    for(std::vector<Gaussian_point>::const_iterator point_it = points.begin();
+        point_it != points.end(); ++point_it) {
+      deviation += pow(point_it->data - point.data, 2);
+    }
+    return sqrt(deviation / points.size());
+  }
   /*Calculates new parameter values for each cluster */
   void calculate_parameters() {
     for(std::vector<Gaussian_center>::iterator center_it = centers.begin();
@@ -239,9 +264,9 @@ protected:
     while(!is_converged && iteration_count++ < maximum_iteration) {
       prev_likelihood = likelihood;
       likelihood = iterate();
-      is_converged = likelihood - prev_likelihood < threshold * likelihood;
+      is_converged = likelihood - prev_likelihood < threshold * fabs(likelihood);
     }
-    //std::cout << likelihood << " " << iteration_count << std::endl;
+    SEG_DEBUG(std::cout << likelihood << " " << iteration_count << std::endl)
     return likelihood;
   }
 
@@ -254,6 +279,7 @@ protected:
       initiate_centers_randomly(number_of_centers);
 
       double likelihood = calculate_fitting();
+      write_center_parameters("center_paramters.txt");
       if(likelihood > max_likelihood) {
         max_centers = centers;
         max_likelihood = likelihood;
@@ -261,6 +287,28 @@ protected:
     }
     centers = max_centers;
   }
+
+  void write_random_centers(const char* file_name,
+                            const std::vector<int>& center_indexes) {
+    std::ofstream output(file_name, std::ios_base::app);
+    for(std::vector<int>::const_iterator it = center_indexes.begin();
+        it != center_indexes.end(); ++it) {
+      output << points[(*it)].data << std::endl;
+    }
+    output.close();
+  }
+
+  void write_center_parameters(const char* file_name) {
+    std::ofstream output(file_name, std::ios_base::app);
+    for(std::vector<Gaussian_center>::iterator center_it = centers.begin();
+        center_it != centers.end(); ++center_it) {
+      output << "mean: " << center_it->mean << " dev: " << center_it->deviation
+             << " mix: " << center_it->mixing_coefficient << std::endl;
+    }
+  }
 };
 }//namespace CGAL
+#ifdef SEG_DEBUG
+#undef SEG_DEBUG
+#endif
 #endif //CGAL_SEGMENTATION_EXPECTATION_MAXIMIZATION_H

Surface_mesh_segmentation/include/CGAL/internal/Surface_mesh_segmentation/K_means_clustering.h

@@ -7,6 +7,13 @@
 #include <cstdlib>
 #include <limits>
 
+//#define ACTIVATE_SEGMENTATION_K_MEANS_DEBUG
+#ifdef ACTIVATE_SEGMENTATION_K_MEANS_DEBUG
+#define SEG_DEBUG(x) x;
+#else
+#define SEG_DEBUG(x)
+#endif
+
 namespace CGAL
 {
 
@@ -84,8 +91,9 @@ public:
     srand(seed);
     calculate_clustering_with_multiple_run(number_of_centers, number_of_run);
   }
-
+  /* For each data point, data_center is filled by its center's id. */
   void fill_with_center_ids(std::vector<int>& data_centers) {
+    data_centers.reserve(points.size());
     for(std::vector<K_means_point>::iterator point_it = points.begin();
         point_it != points.end(); ++point_it) {
       data_centers.push_back(point_it->center_id);
@@ -115,18 +123,21 @@ public:
     bool any_center_changed = true;
     while(any_center_changed && iteration_count++ < maximum_iteration) {
       any_center_changed = false;
+      /* For each point, calculate its new center */
       for(std::vector<K_means_point>::iterator point_it = points.begin();
           point_it != points.end(); ++point_it) {
         bool center_changed = point_it->calculate_new_center(centers);
         any_center_changed |= center_changed;
       }
+      /* For each center, calculate its new mean */
       for(std::vector<K_means_center>::iterator center_it = centers.begin();
           center_it != centers.end(); ++center_it) {
         center_it->calculate_mean();
       }
     }
     is_converged = !any_center_changed;
-    //std::cout << iteration_count << " " << is_converged << std::endl;
+    SEG_DEBUG(std::cout << iteration_count << " " << (is_converged ? "converged" :
+              "not converged") << std::endl)
   }
 
   void calculate_clustering_with_multiple_run(int number_of_centers,
@@ -170,4 +181,7 @@ public:
   }
 };
 }//namespace CGAL
+#ifdef SEG_DEBUG
+#undef SEG_DEBUG
+#endif
 #endif //CGAL_SEGMENTATION_K_MEANS_CLUSTERING_H