{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Histogram APs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "from pathlib import Path\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "triskele_path = Path('../triskele/python/')\n",
    "sys.path.append(str(triskele_path.resolve()))\n",
    "import triskele"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Specific Utils\n",
    "\n",
    "def DFC_filter(raster):\n",
    "    raster[raster > 1e4] = raster[raster < 1e4].max()\n",
    "\n",
    "def show(im, im_size=1, save=None):\n",
    "    plt.figure(figsize=(16*im_size,3*im_size))\n",
    "    plt.imshow(im)\n",
    "    plt.colorbar()\n",
    "    \n",
    "    if save is not None:\n",
    "        plt.savefig(save, bbox_inches='tight', pad_inches=1)\n",
    "    \n",
    "    plt.show()\n",
    "\n",
    "def mshow(Xs, titles=None, im_size=1, save=None):\n",
    "    s = len(Xs)\n",
    "\n",
    "    plt.figure(figsize=(16*im_size,3*im_size*s))\n",
    "\n",
    "    for i in range(s):\n",
    "        plt.subplot(s,1,i+1)\n",
    "        plt.imshow(Xs[i])\n",
    "        \n",
    "        if titles is not None:\n",
    "            plt.title(titles[i])\n",
    "            \n",
    "        plt.colorbar()\n",
    "    \n",
    "    if save is not None:\n",
    "        plt.savefig(save, bbox_inches='tight', pad_inches=1)\n",
    "        \n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load a single raster"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "raster_p = Path('../Data/phase1_rasters/DEM+B_C123/UH17_GEM051_TR.tif')\n",
    "raster   = triskele.read(raster_p)\n",
    "DFC_filter(raster)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "show(raster)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compute attributes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#area = np.array([25, 100, 500, 1e3, 5e3, 10e3, 50e3, 100e3, 150e3])\n",
    "area = [1e3, 1e6]\n",
    "moi  = [.5, .7, .9]\n",
    "sd   = [.1, .5, .9]\n",
    "moi = None\n",
    "sd = None\n",
    "\n",
    "t = triskele.Triskele(raster, verbose=False)\n",
    "attributes = t.filter(tree='tos-tree',\n",
    "                      area=area,\n",
    "                      moment_of_inertia=moi,\n",
    "                      standard_deviation=sd\n",
    "                     )\n",
    "attributes.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "show(attributes[:,:,-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "figs = list()\n",
    "for i in range(attributes.shape[2]):\n",
    "    figs.append(attributes[:,:,i])\n",
    "    \n",
    "mshow(figs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "a = attributes\n",
    "i = 1\n",
    "show(a[:,:,i].astype(np.float) - a[:,:,i+1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compute patches"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "patch_size = 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "\n",
    "offset_stack = None\n",
    "def create_patches(array, patch_size=3):\n",
    "    amp = int((patch_size - 1 ) / 2)\n",
    "\n",
    "    stack = list()\n",
    "    for i in range(-amp, amp+1):\n",
    "        ai = i if i > 0 else None\n",
    "        bi = i if i < 0 else None\n",
    "        ci = -bi if bi is not None else None\n",
    "        di = -ai if ai is not None else None\n",
    "\n",
    "        for j in range(-amp, amp+1):\n",
    "            offset = np.zeros_like(array)\n",
    "            #offset = np.empty(array.shape)\n",
    "            aj = j if j > 0 else None\n",
    "            bj = j if j < 0 else None\n",
    "            cj = -bj if bj is not None else None\n",
    "            dj = -aj if aj is not None else None\n",
    "            print('{}:{} {}:{} - {}:{} {}:{}'.format(ai, bi, ci, di, aj, bj, cj, dj))\n",
    "            offset[ai:bi, aj:bj] = array[ci:di, cj:dj]\n",
    "            stack.append(offset)\n",
    "    return np.stack(stack, axis=-1)\n",
    "\n",
    "offset_stack = create_patches(attributes, patch_size)\n",
    "offset_stack.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " init | user | sys | total\n",
    " --- | ---: | ---: | ---:\n",
    " empty | 27.9 s | 15.2 s | 43.1 s\n",
    " zeros | 27.2 s | 14.1 s | 41.3 s\n",
    " zeros_like | 6.47 s | 4.01 s | 10.5 s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%whos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "offset_stack[-2,-3,:].reshape(-1,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "stack_std = np.std(offset_stack, axis=-1)\n",
    "stack_std.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "figs = list()\n",
    "ttls = list()\n",
    "for d in range(stack_std.shape[-1]):\n",
    "    if d == 0:\n",
    "        ttls.append('Origin')\n",
    "    else:\n",
    "        ttls.append('STD 5x5 from area {}'.format(area[d-1]))\n",
    "    figs.append(stack_std[:,:,d])\n",
    "    \n",
    "mshow(figs, ttls, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "stack_mean = np.mean(offset_stack, axis=-1)\n",
    "stack_avr  = np.average(offset_stack, axis=-1)\n",
    "stack_var  = np.var(offset_stack, axis=-1)\n",
    "stack_min  = np.min(offset_stack, axis=-1)\n",
    "stack_max  = np.max(offset_stack, axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.std.__name__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "create_patches.__name__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for func in (np.mean, np.var):\n",
    "    tmp = func(offset_stack, axis=-1)\n",
    "    show(tmp[:,:,0])\n",
    "    print(tmp.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "size = '{}x{}'.format(patch_size, patch_size)\n",
    "figs = [raster,\n",
    "        stack_avr,\n",
    "        stack_mean,\n",
    "        stack_min,\n",
    "        stack_max,\n",
    "        stack_var,\n",
    "        stack_std\n",
    "       ]\n",
    "ttls = ['Origin',\n",
    "        'Average ' + size,\n",
    "        'Mean ' + size,\n",
    "        'Minimum ' + size,\n",
    "        'Maximum ' + size,\n",
    "        'Variance ' + size,\n",
    "        'STD ' + size\n",
    "       ]\n",
    "mshow(figs, ttls)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "size = '{}x{}'.format(patch_size, patch_size)\n",
    "figs = [raster,\n",
    "        stack_avr[:,:,-1],\n",
    "        stack_mean[:,:,-1],\n",
    "        stack_min[:,:,-1],\n",
    "        stack_max[:,:,-1],\n",
    "        stack_var[:,:,-1],\n",
    "        stack_std[:,:,-1]\n",
    "       ]\n",
    "ttls = ['Origin',\n",
    "        'Average ' + size,\n",
    "        'Mean ' + size,\n",
    "        'Minimum ' + size,\n",
    "        'Maximum ' + size,\n",
    "        'Variance ' + size,\n",
    "        'STD ' + size\n",
    "       ]\n",
    "mshow(figs, ttls, 2, save='mstack2.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### TMP TRISLEK DBG"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "normal = triskele.read('../triskele/build/out/test-default.tif')\n",
    "sd     = triskele.read('../triskele/build/out/test-sd.tif')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "normal.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "areas = np.loadtxt('../triskele/data/areaThresholds.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "figs = [normal[:,:,0]]\n",
    "ttls = ['Origin']\n",
    "\n",
    "for area, i in zip(areas, range(1,normal.shape[2])):\n",
    "    figs.append(normal[:,:,i])\n",
    "    ttls.append('SDAP area {}'.format(area))\n",
    "\n",
    "mshow(figs, ttls, 2, '../Res/trkldbg_triskele_ap.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "figs = [sd[:,:,0]]\n",
    "ttls = ['Origin']\n",
    "\n",
    "for area, i in zip(areas, range(1,sd.shape[2])):\n",
    "    figs.append(sd[:,:,i])\n",
    "    ttls.append('Triskele LFSDAP (SD) area {}'.format(area))\n",
    "\n",
    "mshow(figs, ttls, 2, '../Res/trkldbg_triskele_sd.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "(normal == sd).all()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "offset_stack = create_patches(normal, 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "stack_std = np.std(offset_stack, axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "figs = [stack_std[:,:,0]]\n",
    "ttls = ['Origin std 5x5']\n",
    "\n",
    "for area, i in zip(areas, range(1,sd.shape[2])):\n",
    "    figs.append(stack_std[:,:,i])\n",
    "    ttls.append('LD2DAPs LFSDAP std 5x5 area {}'.format(area))\n",
    "\n",
    "mshow(figs, ttls, 2, '../Res/trkldbg_ld2dap_sd.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "figs = list()\n",
    "ttls = list()\n",
    "for d in range(stack_std.shape[-1]):\n",
    "    if d == 0:\n",
    "        ttls.append('Origin')\n",
    "    else:\n",
    "        ttls.append('STD 5x5 from area {}'.format(area[d-1]))\n",
    "    figs.append(stack_std[:,:,d])\n",
    "    \n",
    "mshow(figs, ttls, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert False"
   ]
  },
  {
   "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"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}