Datapoint-filter-and-visualisation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "719b7b6b-d0b4-4c03-b218-ab7667e1586d",
   "metadata": {},
   "outputs": [
    {
     "ename": "error",
     "evalue": "OpenCV(4.10.0) D:\\a\\opencv-python\\opencv-python\\opencv\\modules\\imgproc\\src\\color.cpp:196: error: (-215:Assertion failed) !_src.empty() in function 'cv::cvtColor'\n",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31merror\u001b[0m                                     Traceback (most recent call last)",
      "Input \u001b[1;32mIn [1]\u001b[0m, in \u001b[0;36m<cell line: 22>\u001b[1;34m()\u001b[0m\n\u001b[0;32m     19\u001b[0m original_img \u001b[38;5;241m=\u001b[39m cv2\u001b[38;5;241m.\u001b[39mimread(original_image_path, cv2\u001b[38;5;241m.\u001b[39mIMREAD_COLOR)\n\u001b[0;32m     21\u001b[0m \u001b[38;5;66;03m# Convert to grayscale and apply Gaussian blur\u001b[39;00m\n\u001b[1;32m---> 22\u001b[0m gray_img \u001b[38;5;241m=\u001b[39m \u001b[43mcv2\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcvtColor\u001b[49m\u001b[43m(\u001b[49m\u001b[43moriginal_img\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mcv2\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mCOLOR_BGR2GRAY\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m     23\u001b[0m blurred_img \u001b[38;5;241m=\u001b[39m cv2\u001b[38;5;241m.\u001b[39mGaussianBlur(gray_img, (\u001b[38;5;241m9\u001b[39m, \u001b[38;5;241m9\u001b[39m), \u001b[38;5;241m0\u001b[39m)\n\u001b[0;32m     25\u001b[0m \u001b[38;5;66;03m# Adaptive thresholding\u001b[39;00m\n",
      "\u001b[1;31merror\u001b[0m: OpenCV(4.10.0) D:\\a\\opencv-python\\opencv-python\\opencv\\modules\\imgproc\\src\\color.cpp:196: error: (-215:Assertion failed) !_src.empty() in function 'cv::cvtColor'\n"
     ]
    }
   ],
   "source": [
    "import cv2\n",
    "import numpy as np\n",
    "from sklearn.cluster import DBSCAN\n",
    "import matplotlib.pyplot as plt\n",
    "from prettytable import PrettyTable\n",
    "import csv\n",
    "\n",
    "# Function to calculate the minimum distance to the nearest neighbor for each centroid\n",
    "def calculate_min_distances(centroids):\n",
    "    min_distances = np.zeros(len(centroids))\n",
    "    for i, center in enumerate(centroids):\n",
    "        distances_to_others = np.sqrt((centroids - center)**2).sum(axis=1)\n",
    "        sorted_distances = np.sort(distances_to_others)\n",
    "        min_distances[i] = sorted_distances[1] if len(sorted_distances) > 1 else 0\n",
    "    return min_distances\n",
    "\n",
    "# Load the original image\n",
    "original_image_path = #enter image path here\n",
    "original_img = cv2.imread(original_image_path, cv2.IMREAD_COLOR)\n",
    "\n",
    "# Convert to grayscale and apply Gaussian blur\n",
    "gray_img = cv2.cvtColor(original_img, cv2.COLOR_BGR2GRAY)\n",
    "blurred_img = cv2.GaussianBlur(gray_img, (9, 9), 0)\n",
    "\n",
    "# Adaptive thresholding\n",
    "thresh_img = cv2.adaptiveThreshold(blurred_img, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, \n",
    "                                   cv2.THRESH_BINARY, 11, 2)\n",
    "\n",
    "# Find contours and filter out small ones\n",
    "contours, _ = cv2.findContours(thresh_img, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)\n",
    "filtered_contours = [c for c in contours if cv2.contourArea(c) > 1]\n",
    "\n",
    "# Calculate centroids\n",
    "centroids = np.array([cv2.moments(c) for c in filtered_contours])\n",
    "centroid_points = np.array([(int(M['m10']/M['m00']), int(M['m01']/M['m00'])) \n",
    "                            if M['m00'] != 0 else (0, 0) for M in centroids])\n",
    "\n",
    "# Save image after blurring and clustering\n",
    "image_with_points = original_img.copy()\n",
    "for point in centroid_points:\n",
    "    cv2.circle(image_with_points, point, 3, (0, 255, 0), -1)\n",
    "intermediate_output_path = 'C:/Users/tnerger/Desktop/Nerv/1.jpg'\n",
    "cv2.imwrite(intermediate_output_path, image_with_points)\n",
    "\n",
    "# Apply DBSCAN clustering to centroids to filter out outliers--------------------------------------------------------------\n",
    "dbscan_eps = 27  # Adjust this value if needed\n",
    "db = DBSCAN(eps=dbscan_eps, min_samples=11).fit(centroid_points)\n",
    "labels = db.labels_\n",
    "largest_cluster_label = max(set(labels), key=list(labels).count)\n",
    "\n",
    "# Create a mask to draw the circles for the largest cluster\n",
    "mask_with_circles = np.zeros(original_img.shape[:2], dtype=\"uint8\")\n",
    "centroids_without_outliers = centroid_points[labels == largest_cluster_label]\n",
    "\n",
    "# Draw filled circles on the mask for each centroid in the largest cluster\n",
    "for center in centroids_without_outliers:\n",
    "    cv2.circle(mask_with_circles, center, int(dbscan_eps), (255, 255, 255), -1)\n",
    "\n",
    "# Create an RGBA image from the original image\n",
    "rgba_image = cv2.cvtColor(original_img, cv2.COLOR_BGR2BGRA)\n",
    "\n",
    "# Set transparent background where the mask is black\n",
    "rgba_image[mask_with_circles == 0] = [0, 0, 0, 0]\n",
    "\n",
    "# Draw filled circles with red color and full opacity on the RGBA image\n",
    "for center in centroids_without_outliers:\n",
    "    cv2.circle(rgba_image, center, int(dbscan_eps), (0, 0, 255, 255), -1)\n",
    "\n",
    "# Save the final image with a transparent background as a PNG\n",
    "final_output_path_png = #enter path here\n",
    "cv2.imwrite(final_output_path_png, rgba_image)\n",
    "\n",
    "# Create and save image with remaining points after DBSCAN without previous points\n",
    "remaining_points_image = original_img.copy()  # Use original image for drawing\n",
    "remaining_points_image[:] = (0, 0, 0)  # Set image to black\n",
    "\n",
    "for point in centroids_without_outliers:\n",
    "    cv2.circle(remaining_points_image, point, 2, (0, 0, 255), -1)  # Draw red points\n",
    "    cv2.circle(remaining_points_image, point, int(dbscan_eps), (0, 0, 255), 2)  # Draw red circles around the points\n",
    "\n",
    "remaining_points_output_path = #enter path here\n",
    "cv2.imwrite(remaining_points_output_path, remaining_points_image)\n",
    "\n",
    "# Calculate the minimum distances\n",
    "min_distances = calculate_min_distances(centroid_points)\n",
    "\n",
    "# Filter out distances that are greater than 0\n",
    "filtered_distances = [distance for distance in min_distances if distance > 0]\n",
    "\n",
    "# Create a PrettyTable and a list for CSV output\n",
    "table = PrettyTable()\n",
    "table.field_names = [\"Nummer\", \"X-Koordinate\", \"Y-Koordinate\", \"Minimale Distanz\"]\n",
    "\n",
    "csv_data = []\n",
    "\n",
    "# Add each row to the table and CSV data, excluding zero distances\n",
    "num = 1\n",
    "for i, (center, distance) in enumerate(zip(centroids_without_outliers, min_distances)):\n",
    "    if distance > 0:\n",
    "        x, y = center\n",
    "        table.add_row([num, x, y, distance])\n",
    "        csv_data.append([num, x, y, distance])\n",
    "        num += 1\n",
    "\n",
    "# Print the table\n",
    "print(table)\n",
    "\n",
    "# Save the data to a CSV file\n",
    "csv_output_path = #enter path here\n",
    "with open(csv_output_path, mode='w', newline='') as file:\n",
    "    writer = csv.writer(file)\n",
    "    writer.writerow([\"Nummer\", \"X-Koordinate\", \"Y-Koordinate\", \"Minimale Distanz\"])\n",
    "    writer.writerows(csv_data)\n",
    "\n",
    "# Create a histogram of the distances with integer bins starting from 0\n",
    "plt.figure(figsize=(8, 6))  # Reduce the width by 20%\n",
    "bins = np.arange(0, max(filtered_distances) + 2) - 0.5  # Create bins for each integer distance\n",
    "plt.hist(filtered_distances, bins=bins, edgecolor='black', color='red')\n",
    "plt.title('Histogram of Minimum Distances')\n",
    "plt.xlabel('Distance / Pixel')\n",
    "plt.ylabel('Number of Points')\n",
    "plt.xticks(np.arange(0, max(filtered_distances) + 1, 10))  # Set x-ticks for 0, 10, 20, ...\n",
    "plt.grid(axis='y')  # Only horizontal grid lines\n",
    "\n",
    "# Adjust the layout to minimize the distance between the y-axis and 0\n",
    "plt.gca().set_xlim(left=-0.5)\n",
    "\n",
    "plt.savefig(#enter path here)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f0666b77-8a1a-44b4-b88a-64ac2d70cbb1",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}