project1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import cv2\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "import itertools\n",
    "import pandas as pd\n",
    "import PIL"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Image segmentation + Resampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1243c6cc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def rescale_segment( segment, size = [28,28] ):\n",
    "    '''function for resizing (scaling down) images\n",
    "    input parameters\n",
    "    seg : the segment of image (np.array)\n",
    "    size : out size (list of two integers)\n",
    "    output \n",
    "    scaled down image'''\n",
    "    if len(segment.shape) == 3 : # Non Binary Image\n",
    "        import cv2\n",
    "        # thresholding the image\n",
    "        ret,segment = cv2.threshold(segment,127,255,cv2.THRESH_BINARY)\n",
    "    m,n = segment.shape\n",
    "    idx1 = list(range(0,m, (m)//(size[0]) ) )\n",
    "    idx2 = list(range(0,n, n//(size[1]) )) \n",
    "    out = np.zeros(size)\n",
    "    for i in range(size[0]):\n",
    "        for j in range(size[1]):\n",
    "            out[i,j] = segment[ idx1[i] + (m%size[0])//2, idx2[j] + (n%size[0])//2]\n",
    "    kernel = np.ones( [2,2])\n",
    "    return out\n",
    "\n",
    "def extract_segments(img, pad=30, reshape = 0,size = [28,28]) :\n",
    "    '''function to extract individual chacters and digits from an image\n",
    "    input paramterts\n",
    "    img : input image (numpy array)\n",
    "    pad : padding window size around segments (int)\n",
    "    size : out size (list of two integers)\n",
    "    reshape : if 1 , output will be scaled down. if 0, no scaling down\n",
    "    Returns\n",
    "    out : list of each segments (starting from leftmost digit)'''\n",
    "    \n",
    "    import cv2\n",
    "    # thresholding the image\n",
    "    ret,thresh1 = cv2.threshold(img,127,255,cv2.THRESH_BINARY)\n",
    "    \n",
    "    # Negative tranform gray levels (background becomes black) \n",
    "    thresh1 = thresh1 - 1\n",
    "    thresh1[thresh1 == 254] = 0\n",
    "    thresh1[thresh1 == - 1 ] = 255\n",
    "    \n",
    "    # connected component labelling \n",
    "    output = cv2.connectedComponentsWithStats(thresh1, 4)\n",
    "    final = []\n",
    "    temp1 = np.sort( output[2][:,0] )\n",
    "    kernel = np.ones( [3,3])\n",
    "    for i in range(1,output[0]):\n",
    "        temp2 = output[2]\n",
    "        cord = np.squeeze( temp2[temp2[:,0] == temp1[i]] )\n",
    "        num = np.pad( thresh1[ cord[1]:cord[1]+cord[3], cord[0]:cord[0]+cord[2] ], pad,'constant')\n",
    "        num = cv2.dilate(num,kernel,iterations = 1)\n",
    "        if reshape == 1:\n",
    "            num = rescale_segment( num, size )\n",
    "        final.append(num)\n",
    "        \n",
    "    return final\n",
    "\n",
    "# reading the image,  0 would convert to gray image\n",
    "img = cv2.imread('/Users/josejoy/Desktop/ECE 271B Stat Learning /project/eq1.jpg',0)\n",
    "segments= extract_segments(img, 30, reshape = 1, size = [28,28] )\n",
    "plt.imshow(segments[0],cmap = 'gray')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Morphological operations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nu66mdsZKAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11f94d748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "kernel = np.ones( [2,2])\n",
    "opening = cv2.morphologyEx(segments[0], cv2.MORPH_CLOSE, kernel )\n",
    "# dilation = cv2.dilate(resam,kernel,iterations = 1)\n",
    "plt.imshow(opening,cmap = 'gray')\n",
    "plt.show()  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## HASy dataset manipulation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "| Symbols | symbol_id |\n",
    "|:---------|-----------|:-\n",
    "| pi| 81|\n",
    "| < | 187|\n",
    "|>|188|\n",
    "|-|195|\n",
    "|+|196|\n",
    "|%|267|\n",
    "| division | 526 |\n",
    "| multiply | 513 |\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>path</th>\n",
       "      <th>symbol_id</th>\n",
       "      <th>latex</th>\n",
       "      <th>user_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>../../hasy-data/v2-00016.png</td>\n",
       "      <td>31</td>\n",
       "      <td>A</td>\n",
       "      <td>8071</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>../../hasy-data/v2-00017.png</td>\n",
       "      <td>31</td>\n",
       "      <td>A</td>\n",
       "      <td>9505</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>../../hasy-data/v2-00018.png</td>\n",
       "      <td>31</td>\n",
       "      <td>A</td>\n",
       "      <td>16925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>../../hasy-data/v2-00019.png</td>\n",
       "      <td>31</td>\n",
       "      <td>A</td>\n",
       "      <td>16925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>../../hasy-data/v2-00020.png</td>\n",
       "      <td>31</td>\n",
       "      <td>A</td>\n",
       "      <td>16925</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                           path  symbol_id latex  user_id\n",
       "0  ../../hasy-data/v2-00016.png         31     A     8071\n",
       "1  ../../hasy-data/v2-00017.png         31     A     9505\n",
       "2  ../../hasy-data/v2-00018.png         31     A    16925\n",
       "3  ../../hasy-data/v2-00019.png         31     A    16925\n",
       "4  ../../hasy-data/v2-00020.png         31     A    16925"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "path1 = '/Users/josejoy/Desktop/ECE 271B Stat Learning /project/HASYv2/'\n",
    "symbols = pd.read_csv(path1+'symbols.csv')\n",
    "path2 = '/Users/josejoy/Desktop/ECE 271B Stat Learning /project/HASYv2/classification-task/fold-1/'\n",
    "train1 = pd.read_csv(path2+'train.csv')\n",
    "train1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKsr4paKMXyrq1yZ2fSeaWQvEAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x125abc208>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fname1 = train1['path'][1]\n",
    "path3 = '/Users/josejoy/Desktop/ECE 271B Stat Learning /project/HASYv2/hasy-data/'\n",
    "temp = train1[train1['symbol_id']== 81 ]\n",
    "name1 = temp.iloc[11]['path']\n",
    "def get_Hasy_filename(name):\n",
    "    '''function to extract filenames for reading images from HASy dataset\n",
    "    Input : name, string of file path\n",
    "    Output : only file name'''\n",
    "    for i in range(1,len(name)):\n",
    "        if name[-i] == 'v':\n",
    "            return name[-i:]\n",
    "        \n",
    "img1 = cv2.imread(path3 + get_Hasy_filename(name1),0)\n",
    "img1 = rescale_segment( img1 )\n",
    "# img1 = PIL.Image.open(path3 + 'v2-25562.png')\n",
    "plt.imshow(img1)\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}