Character Segmentation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "def remove_dots(img, baseline, th, file_name): #th is the minimum number of pixels for it to not be a dot or hamza\n",
    "    desired_lower_region_size = 9\n",
    "    desired_higher_region_size = 15\n",
    "    top_region_size = baseline if baseline - desired_higher_region_size < 0 else desired_higher_region_size\n",
    "    bot_region_size = img.shape[0] - baseline + 1 if baseline + desired_lower_region_size + 1 > img.shape[0] else desired_lower_region_size\n",
    "    cropped_img = img[baseline - top_region_size: baseline + bot_region_size, :]\n",
    "    cv2.imwrite(file_name + '-cropped.png', cropped_img)\n",
    "    visited = np.zeros(cropped_img.shape)\n",
    "    for row in range(cropped_img.shape[0]):\n",
    "        for col in range(cropped_img.shape[1]):\n",
    "            if (cropped_img[row, col] != 255 or visited[row, col]):\n",
    "                continue\n",
    "            visited_copy = np.copy(visited)\n",
    "            count = dfs(cropped_img, row, col, visited, False)\n",
    "            print(count)\n",
    "            if count <= th:\n",
    "                _ = dfs(cropped_img, row, col, visited_copy, True)\n",
    "                img[baseline - top_region_size: baseline + bot_region_size, :] = cropped_img\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {},
   "outputs": [],
   "source": [
    "from math import pi, sqrt, exp\n",
    "\n",
    "def gauss(n=11,sigma=1):\n",
    "    r = range(-int(n/2),int(n/2)+1)\n",
    "    return [1 / (sigma * sqrt(2*pi)) * exp(-float(x)**2/(2*sigma**2)) for x in r]\n",
    "\n",
    "def convolve(img, kernel):\n",
    "    result = np.zeros(len(img))\n",
    "    half_width = len(kernel) // 2\n",
    "    for i in range(half_width, len(img) - half_width):\n",
    "        val = 0\n",
    "        for j in range(-half_width, half_width + 1):\n",
    "            val += img[i + j] * kernel [j + half_width]\n",
    "        result[i] = val\n",
    "    return result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2\n",
    "import numpy as np\n",
    "import numpy.ma as ma\n",
    "import os, glob, re\n",
    "import shutil\n",
    "from scipy.fftpack import dct, idct\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "#from scipy import ndimage\n",
    "numbers = re.compile(r'(\\d+)')\n",
    "def numericalSort(value):\n",
    "    parts = numbers.split(value)\n",
    "    parts[1::2] = map(int, parts[1::2])\n",
    "    return parts\n",
    "results_path = 'C:/D/PDFS/College/4th year/1st term/Pattern Recognition/Project/Dataset/results'\n",
    "char_path = 'C:/D/PDFS/College/4th year/1st term/Pattern Recognition/Project/Dataset/chars'\n",
    "char_folder_path = 'C:/D/PDFS/College/4th year/1st term/Pattern Recognition/Project/Dataset/'\n",
    "\n",
    "dx = [-1, 1, 0, 0, 1, -1, 1, -1]\n",
    "dy = [0, 0, -1, 1, 1, -1, -1, 1]\n",
    "#dx = [-1, 1, 0, 0]\n",
    "#dy = [0, 0, -1, 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [],
   "source": [
    "def is_valid(row, col, img_shape):\n",
    "    if row < 0 or row >= img_shape[0] or col < 0 or col >= img_shape[1]:\n",
    "        return False\n",
    "    return True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {},
   "outputs": [],
   "source": [
    "def dfs(img, row, col, visited, remove):\n",
    "    visited[row, col] = True\n",
    "    count = 1\n",
    "    for i in range(len(dx)):\n",
    "        new_row = row + dx[i]\n",
    "        new_col = col + dy[i]\n",
    "        if is_valid(new_row, new_col, img.shape) and not visited[new_row, new_col] and img[new_row, new_col] == 255:\n",
    "            new_count = dfs(img, new_row, new_col, visited, remove)\n",
    "            count += new_count\n",
    "    if remove:\n",
    "        img[row, col] = 0\n",
    "    return count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_dots(img, baseline, th, file_name): #th is the minimum number of pixels for it to not be a dot or hamza\n",
    "    visited = np.zeros(img.shape, dtype=bool)\n",
    "    for col in range(img.shape[1]):\n",
    "        if (img[baseline, col] != 255 or visited[baseline, col]):\n",
    "            continue\n",
    "        visited_copy = np.copy(visited)\n",
    "        count = dfs(img, baseline, col, visited, False)\n",
    "        if count <= th:\n",
    "            _ = dfs(img, baseline, col, visited_copy, True)\n",
    "    #cv2.imwrite(file_name+\"before.png\", img)\n",
    "    img = img == 255\n",
    "    return (np.logical_and(img, visited).astype('uint8')) * 255\n",
    "    #cv2.imwrite(file_name+'after.png', img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {},
   "outputs": [],
   "source": [
    "def segment_characters(img_name, folder_count, img_count):\n",
    "    file_name = char_path + '/' + str(folder) + '/result' + str(img_count) + '/' + 'result' + str(folder_count) + '-' + str(img_count)\n",
    "    img = cv2.imread(img_name, 0)  #0 parameter means read it in grayscale\n",
    "    _, img = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY_INV|cv2.THRESH_OTSU)\n",
    "    img_copy = img.copy()\n",
    "    horizontal_profile = cv2.reduce(img, 1, cv2.REDUCE_SUM, dtype=cv2.CV_32S) #2nd param = 1 means row sum\n",
    "    baseline_idx = np.argmax(horizontal_profile)\n",
    "    dot_count_threshold = 6\n",
    "    no_dots_img = remove_dots(img_copy, baseline_idx, dot_count_threshold, file_name)\n",
    "    #no_dots_img = 255 - no_dots_img\n",
    "    cv2.imwrite(file_name + \"_no_dots.png\", no_dots_img)\n",
    "    vertical_profile = cv2.reduce(no_dots_img, 0, cv2.REDUCE_SUM, dtype=cv2.CV_32S)\n",
    "    #print(vertical_profile)\n",
    "    #result = convolve(vertical_profile[0], gauss(3, 0.1))\n",
    "    #result = convolve(vertical_profile[0], [-1, -2, 0, 2, 1])\n",
    "    #print(np.fft.fft(vertical_profile)[0][0:4])\n",
    "    #print(vertical_profile[0][10:30])\n",
    "    #idk = idct(dct(vertical_profile)[0][0:2], 2, n=len(vertical_profile[0]))\n",
    "    #result = convolve(idk, [-1, 0, 1])\n",
    "    results = np.abs(convolve(vertical_profile[0], [-1, 0, 1]))\n",
    "    mask = np.zeros(len(results) - 1, dtype=bool)\n",
    "    #print(result)\n",
    "    mask = results <= 255\n",
    "    for i in range(len(results) - 1):\n",
    "        if mask[i] == True and mask[i+1] == True:\n",
    "            mask[i] = False\n",
    "    #print(result[mask])\n",
    "    '''\n",
    "    for i in range(len(vertical_profile[0]) - 1):\n",
    "        if vertical_profile[0][i] == vertical_profile[0][i + 1]:\n",
    "            cv2.line(img, (i, 0), (i, img.shape[0]), (255,0,0), 1)\n",
    "    '''\n",
    "    verticals = np.argwhere(mask == True)\n",
    "    for x in verticals:\n",
    "        cv2.line(img, (x[0], 0), (x[0], img.shape[0]), (255,0,0), 1)\n",
    "    #print('result of convolution:')\n",
    "    #print(result)\n",
    "    cv2.imwrite(file_name + \".png\", img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result1\n",
      "result0.png\n"
     ]
    },
    {
     "ename": "NameError",
     "evalue": "name 'results' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-262-92bb6de60fe1>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m     14\u001b[0m         \u001b[0mos\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mmkdir\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mchar_path\u001b[0m \u001b[1;33m+\u001b[0m \u001b[1;34m'/'\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mfolder\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m+\u001b[0m \u001b[1;34m'/result'\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mimg_count\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     15\u001b[0m         \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mimg_name\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 16\u001b[1;33m         \u001b[0msegment_characters\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mimg_name\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfolder_count\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mimg_count\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     17\u001b[0m     \u001b[0mos\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mchdir\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'..'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-261-fdec21c688a1>\u001b[0m in \u001b[0;36msegment_characters\u001b[1;34m(img_name, folder_count, img_count)\u001b[0m\n\u001b[0;32m     23\u001b[0m     \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mplot\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mresults1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     24\u001b[0m     \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mplot\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mresults2\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 25\u001b[1;33m     \u001b[0mmask\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mzeros\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mresults\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m-\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mbool\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     26\u001b[0m     \u001b[1;31m#print(result)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     27\u001b[0m     \u001b[0mmask\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mresults1\u001b[0m \u001b[1;33m<=\u001b[0m \u001b[1;36m255\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m: name 'results' is not defined"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "if os.path.exists(char_folder_path + 'chars'):\n",
    "    shutil.rmtree(char_folder_path + 'chars', ignore_errors=True)\n",
    "os.mkdir(char_folder_path + 'chars')\n",
    "os.chdir(results_path)\n",
    "for folder_count, folder in enumerate(os.listdir()):\n",
    "    if (folder_count >= 1):\n",
    "        break\n",
    "    os.mkdir(char_path + '/' + str(folder))\n",
    "    print(folder)\n",
    "    os.chdir(folder)\n",
    "    for img_count, img_name in enumerate(sorted(os.listdir(), key=numericalSort)):\n",
    "        if (img_count >= 20):\n",
    "            break\n",
    "        os.mkdir(char_path + '/' + str(folder) + '/result' + str(img_count))\n",
    "        print(img_name)\n",
    "        segment_characters(img_name, folder_count, img_count)\n",
    "    os.chdir('..')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {},
   "outputs": [],
   "source": [
    "zz = np.array([[False ,True], [False, False]])\n",
    "bb = np.array([[255, 255], [255, 255]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [],
   "source": [
    "def zzz(z, b):\n",
    "    return ma.masked_array(b, z).data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[255, 255],\n",
       "       [255, 255]])"
      ]
     },
     "execution_count": 212,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "zzz(zz, bb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[510.+0.j,   0.+0.j],\n",
       "       [510.+0.j,   0.+0.j]])"
      ]
     },
     "execution_count": 257,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.fft.fft(bb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}