added DBScan simple example

Author: lucamors

DBScan/main.c

@@ -0,0 +1,163 @@
+#include <SDL.h>
+#include <time.h>
+#include <stdio.h>
+#include <stdbool.h> 
+
+#include "gui.h"
+#include "point.h"
+#include "dbscan.h"
+#include "gaussian_distribution.h"
+
+int main()
+{
+    // Init SDL
+    ////////////////////////////////////////////////////////////
+
+    // Define Window Width & Height
+    const int WINDOW_WIDTH  = 600;
+    const int WINDOW_HEIGHT = 600;
+
+    // Create window and renderer
+    SDL_Window* window     = create_window(WINDOW_WIDTH, WINDOW_HEIGHT);
+    SDL_Renderer* renderer = create_renderer(&window);
+
+    // Enable alpha blending
+    SDL_SetRenderDrawBlendMode(renderer, SDL_BLENDMODE_BLEND);
+
+    // Set window color to black
+    clear_window(&renderer);
+  
+    ////////////////////////////////////////////////////////////
+    // Generate k gaussian clusters dataset
+    ////////////////////////////////////////////////////////////
+
+    // Init random number generator
+    srand(time(NULL));
+
+    const int N = 5000;
+    Point dataset[N];
+    
+    const int k = 5;
+    GaussianDistribution gaussian_distributions[k];
+
+    for(int i = 0; i < k; i++)
+    {
+        gaussian_distributions[i].mu_x    = ((double)rand())/RAND_MAX * WINDOW_WIDTH*0.8 + WINDOW_WIDTH*0.1;
+        gaussian_distributions[i].mu_y    = ((double)rand())/RAND_MAX * WINDOW_HEIGHT*0.8 + WINDOW_HEIGHT*0.1;
+        gaussian_distributions[i].sigma_x = ((double)rand())/RAND_MAX * 40 + 10;
+        gaussian_distributions[i].sigma_y = ((double)rand())/RAND_MAX * 40 + 10;
+    }
+    
+    generate_gaussian_clusters_dataset(dataset, N, gaussian_distributions, k);
+   
+    draw_classified_dataset(dataset, N, &renderer);
+    SDL_RenderPresent(renderer);
+
+    ///////////////////////////////////////////////////////////////
+    // Density Based Spatial Clustering of applications with Noise
+    // (DBSCAN) Algorithm
+    ///////////////////////////////////////////////////////////////
+
+    // Algorithm hyper-parameters
+    double eps = 5;
+    double minPts = 10;
+
+    int cluster_counter = 0;
+
+    for(int i = 0; i < N; i++)
+    {
+        
+        // Check if point was already processed
+        // in the inner loop
+        if(dataset[i].label != -1)
+        {
+            continue;
+        }   
+
+        Vector* neighbors = find_neighbors(dataset,N, dataset[i], eps, BRUTE);
+        
+        if( neighbors->size < minPts )
+        {
+            dataset[i].label = -2;
+            free(neighbors->array);
+            free(neighbors);
+            continue;
+        }
+        
+        cluster_counter++;
+
+        // Assign to point the current cluster number
+        dataset[i].label = cluster_counter;
+
+
+        for(int j = 0; j < neighbors->size; j++)
+        {
+            if( neighbors->array[j]->x == dataset[i].x & neighbors->array[j]->y == dataset[i].y)
+            {
+                continue;
+            }
+
+            if(j%10==0)
+            {
+                clear_window(&renderer);
+                draw_classified_dataset(dataset, N, &renderer);
+                SDL_RenderPresent(renderer);
+            }
+            
+
+            if(neighbors->array[j]->label == -2)
+            {
+                neighbors->array[j]->label = cluster_counter;
+            }
+
+            if(neighbors->array[j]->label != -1)
+            {
+                continue;
+            }
+
+            neighbors->array[j]->label = cluster_counter;
+
+            Vector * n2 = find_neighbors(dataset, N, *neighbors->array[j], eps, BRUTE);
+
+            for(int k = 0; k < n2->size; k++)
+            {
+                vector_push_back(neighbors, n2->array[k]);
+            }
+
+            free(n2->array);
+            free(n2);    
+        }
+        
+        // free dynamic neighbors array
+        free(neighbors->array);
+        free(neighbors);
+        
+    }
+
+    printf("Found %d clusters\n",cluster_counter);
+
+    clear_window(&renderer);
+    SDL_SetRenderDrawColor(renderer, 0, 0, 0, 255); // RGBA
+    SDL_RenderClear(renderer);
+    draw_classified_dataset(dataset, N, &renderer);
+    SDL_RenderPresent(renderer);
+    
+     // SDL Visualization Loop
+    SDL_Event e;
+    bool quit = false;
+    while (!quit) 
+    {
+
+        while (SDL_PollEvent(&e) != 0)
+        {
+            
+            if (e.type == SDL_QUIT) 
+            {
+                quit = 1;
+            }
+        }
+    }
+
+
+    return 0;
+}