{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Working with images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib notebook\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import imageio.v2 as imageio"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Image loading and display"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "image = imageio.imread('taj_mahal_gray.jpg')\n",
    "\n",
    "start, step = 300, 30\n",
    "print(f'The image \\n{image}\\nhas {image.shape[0]} rows and {image.shape[1]} columns.')\n",
    "print(f'The {step} pixels in the zero-th row, starting at column {start} contain the values \\n{image[0, start:start + step]}.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "_, ax = plt.subplots(1, 2, figsize=(9, 3))\n",
    "ax[0].imshow(image)\n",
    "ax[1].imshow(image, cmap='gray')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Deep and shallow copy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "image = imageio.imread('taj_mahal_gray.jpg')\n",
    "\n",
    "# Assign another name to the same underlying storage (\"Reference\")\n",
    "# copy = image\n",
    "\n",
    "# Copy the underlying data to a new storage location (\"Copy\")\n",
    "copy = image.copy()\n",
    "\n",
    "copy[200:350, 300:450] = 0\n",
    "\n",
    "fig, ax = plt.subplots(1, 2, figsize=(9, 3))\n",
    "ax[0].imshow(image)\n",
    "ax[1].imshow(copy, cmap='gray')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Image statistics & manipulations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "im_integral = imageio.imread('taj_mahal_gray.jpg')\n",
    "im_float = im_integral / 255.  # normalize grayvalues to [0,1]\n",
    "\n",
    "fig, axs = plt.subplots(1, 2, figsize=(9, 3))\n",
    "for ax, im in zip(axs, [im_integral, im_float]):\n",
    "    artist = ax.imshow(im, cmap='gray')\n",
    "    ax.set_title(f'min={im.min()}, max={im.max()}')\n",
    "    fig.colorbar(artist, ax=ax)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "im = imageio.imread('taj_mahal_small.jpg')\n",
    "print(im.shape)\n",
    "_, ax = plt.subplots()\n",
    "ax.imshow(im)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Conversion to grayscale image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "im = im / 255.0\n",
    "\n",
    "coeffs = np.array([0.299, 0.587, 0.114])\n",
    "gray_explicit = im[:,:,0] * 0.299 + im[:,:,1] * 0.587 + im[:,:,2] * 0.114\n",
    "gray_broadcasting = (im * coeffs).sum(axis=2)\n",
    "assert np.allclose(gray_broadcasting, gray_explicit)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "gray_mean = im.mean(-1)\n",
    "\n",
    "_, ax = plt.subplots(1, 2, figsize=(9, 3))\n",
    "ax[0].imshow(gray_broadcasting, cmap='gray')\n",
    "ax[1].imshow(gray_mean, cmap='gray')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Image filtering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import scipy.ndimage as sn\n",
    "from skimage.transform import rescale"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load image\n",
    "im = imageio.imread('taj_mahal_gray.jpg') / 255.0\n",
    "im_aa = rescale(im, 0.2, anti_aliasing=True)\n",
    "im_noaa = rescale(im, 0.2, anti_aliasing=False)\n",
    "\n",
    "_, ax = plt.subplots(1, 2, figsize=(9,3))\n",
    "ax[0].imshow(im_aa, cmap='gray')\n",
    "ax[0].set_title('AA')\n",
    "ax[1].imshow(im_noaa, cmap='gray')\n",
    "ax[1].set_title('No AA')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Gaussian filter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sigma = 2.3\n",
    "kernel_size = 15\n",
    "\n",
    "space = np.arange(-kernel_size // 2 + 1, kernel_size // 2 + 1)\n",
    "x, y = np.meshgrid(space, space)\n",
    "\n",
    "kernel = np.exp(-(x ** 2 + y ** 2) / (2 * sigma ** 2))\n",
    "kernel = kernel / np.sum(kernel)\n",
    "_, ax = plt.subplots(subplot_kw={'projection': '3d'})\n",
    "ax.plot_surface(x, y, kernel, cmap='viridis')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Apply gaussian filter to image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convolution with the gaussian filter\n",
    "filtered = sn.convolve(im, kernel)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "_, ax = plt.subplots(1, 2, figsize=(9,3));\n",
    "ax[0].imshow(im, cmap='gray')\n",
    "ax[0].set_title('Original')\n",
    "ax[1].imshow(filtered, cmap='gray')\n",
    "ax[1].set_title('2D filter')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Edge filter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dx = np.array([\n",
    "    [-1, 0, 1],\n",
    "    [-1, 0, 1],\n",
    "    [-1, 0, 1]\n",
    "])\n",
    "dy = dx.T\n",
    "\n",
    "_, ax = plt.subplots(1, 2, figsize=(5,2))\n",
    "ax[0].imshow(dx, cmap='gray')\n",
    "ax[1].imshow(dy, cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "nabla_x = sn.convolve(im, dx)\n",
    "nabla_y = sn.convolve(im, dy)\n",
    "grad_magnitude = np.sqrt(nabla_x ** 2 + nabla_y ** 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fi, ax = plt.subplots(1, 3, figsize=(9,3))\n",
    "ax[0].imshow(nabla_x, cmap='gray')\n",
    "ax[1].imshow(nabla_y, cmap='gray')\n",
    "ax[2].imshow(grad_magnitude, cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "nabla_x = sn.convolve(filtered, dx)\n",
    "nabla_y = sn.convolve(filtered, dy)\n",
    "grad_magnitude = np.sqrt(nabla_x ** 2 + nabla_x ** 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fi, ax = plt.subplots(1, 3, figsize=(9,3))\n",
    "ax[0].imshow(nabla_x, cmap='gray')\n",
    "ax[1].imshow(nabla_y, cmap='gray')\n",
    "ax[2].imshow(grad_magnitude, cmap='gray')"
   ]
  }
 ],
 "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": "3.10.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}