mirror of
https://github.com/saymrwulf/nvidia-AI.summerschool17-notebooks.git
synced 2026-11-20 09:52:58 +00:00
641 lines
1.4 MiB
Text
641 lines
1.4 MiB
Text
|
|
{
|
||
|
|
"metadata": {
|
||
|
|
"name": "045-labeling"
|
||
|
|
},
|
||
|
|
"nbformat": 3,
|
||
|
|
"nbformat_minor": 0,
|
||
|
|
"worksheets": [
|
||
|
|
{
|
||
|
|
"cells": [
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Title: Connected Component Labeling\n",
|
||
|
|
"Author: Thomas Breuel\n",
|
||
|
|
"Institution: UniKL"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": true,
|
||
|
|
"input": [
|
||
|
|
"\n",
|
||
|
|
"from scipy.ndimage import filters\n",
|
||
|
|
"def ims(image,**kw):\n",
|
||
|
|
" size = kw.get(\"s\",8)\n",
|
||
|
|
" if \"s\" in kw: del kw[\"s\"]\n",
|
||
|
|
" subplots(1,1,figsize=(size,size))\n",
|
||
|
|
" gray(); imshow(image,**kw)\n",
|
||
|
|
"def imp(image,**kw):\n",
|
||
|
|
" subplots(1,1,figsize=(6,6))\n",
|
||
|
|
" gray(); imshow(image,interpolation='nearest',**kw)\n",
|
||
|
|
"def imrow(*args,**kw):\n",
|
||
|
|
" size = kw.get(\"s\",8)\n",
|
||
|
|
" if \"s\" in kw: del kw[\"s\"]\n",
|
||
|
|
" n = len(args)\n",
|
||
|
|
" gray()\n",
|
||
|
|
" subplots(1,n,figsize=(n*size,size))\n",
|
||
|
|
" for i,im in enumerate(args):\n",
|
||
|
|
" subplot(1,n,i+1); imshow(im,**kw)\n",
|
||
|
|
"def rescale(image):\n",
|
||
|
|
" return (image-amin(image))/max(1e-4,amax(image)-amin(image))"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [],
|
||
|
|
"prompt_number": 1
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Connected Component Labeling\n",
|
||
|
|
"============================\n",
|
||
|
|
"\n",
|
||
|
|
"We have been using binary images and morphology.\n",
|
||
|
|
"\n",
|
||
|
|
"At some point, it becomes important to turn the parts\n",
|
||
|
|
"of binary images into \"objects\".\n",
|
||
|
|
"\n",
|
||
|
|
"This is a common operation in artificial intelligence:\n",
|
||
|
|
"\n",
|
||
|
|
"- start with an analog signal\n",
|
||
|
|
"- turn it into a binary signal somehow\n",
|
||
|
|
"- divide the binary signal into objects\n",
|
||
|
|
"- represent each object as a _symbol_\n",
|
||
|
|
"- continue with _symbolic processing_"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"image = mean(imread(\"testpar.png\"),axis=2)\n",
|
||
|
|
"image /= amax(image)\n",
|
||
|
|
"image = 1-image\n",
|
||
|
|
"imrow(image,s=12)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAAAscAAAJKCAYAAADA2OgoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n1bqOHTvGHUvnqFevXh1TU1MR
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 2
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": true,
|
||
|
|
"input": [
|
||
|
|
"from scipy.ndimage import measurements,morphology"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [],
|
||
|
|
"prompt_number": 3
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Labeling is carried out with the `label` function."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"labels,n = measurements.label(image>0.5)\n",
|
||
|
|
"imrow(labels,s=12)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAAAscAAAJKCAYAAADA2OgoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ntuTHLCKkiAcMNU2b2Md9mggE
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 4
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"That's not very illuminating.\n",
|
||
|
|
"Let's instead recolor the gray values pseudo-randomly.\n",
|
||
|
|
"The function `recolor` does this, keeping the value 0 as black.\n",
|
||
|
|
"We use the `spectral` color map, which gives us a wide range\n",
|
||
|
|
"of colors."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"imrow(mgrid[:200,:1000][1]/1000.0,cmap=cm.spectral)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAAAeIAAAB3CAYAAADfNUPnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nbGyM/Px8iouLAY3xgzA0NMTO
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 5
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"def recolor(labels): return where(labels>0,sin(100*labels)**2+0.1,0)\n",
|
||
|
|
"imrow(recolor(labels),cmap=cm.spectral,s=12)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAAAscAAAJKCAYAAADA2OgoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsnWeY1UQXgN+7fRdYlt6bsPTO0kGRIlVFKYqiiICKoCJI\nVTpIEUGafBYERUTsolQBpXeQIr1K722B7fv9mNybSW5u28ICzrvPPEkmk2RuNpmcOXPOGVtycnIy\nCoVCoVAoFAqFAr+MroBCoVAoFAqFQnGvoIRjhUKhUCgUCoVCQwnHCoVCoVAoFAqFhhKOFQqFQqFQ\nKBQKDSUcKxQKhUKhUCgUGko4VigUCoVCoVAoNNJFOF6yZAmlS5cmMjKScePGpcclFAqFQqFQKBSK\nNMeW1nGOExMTKVWqFMuXL6dAgQJUr16defPmUaZMmbS8jEKhUCgUCoVCkeakueZ48+bNlChRgqJF\nixIYGMizzz7Lr7/+mtaXUSgUCoVCoVAo0pyAtD7h6dOnKVSokGO7YMGCbNq0yVDGZrOl9WUVCoVC\noVAoFApLfDGUSHPNsRJ8FQqFQqFQKBT3K2kuHBcoUICTJ086tk+ePEnBggXT+jIKhUKhUCgUCkWa\nk+bCcVRUFIcOHeL48ePExcUxf/58nnjiibS+jEKhUCgUCoVCkeakuc1xQEAA06ZNo2nTpiQmJtKl\nSxcVqUKhUCgUCoVCcV+Q5qHcvLqosktWKBQKhUKhUNwlMtQhT6FQKBQKhUKhuF9RwrFCoVAoFAqF\nQqGhhGOFQqFQKBQKhUJDCccKhUKhUCgUCoWGEo4VCoVCoVAoFAoNJRwrFAqFQqFQKBQaSjhWKBQK\nhUKhUCg00nwSEEUaE/Wovn7kH7h6Qd8uXh6O7IESFeDwboisCFlzwKGdcP0KlKwEpapAm1dF+XE9\nYd+2u1r9YrmhajGx/vt2iI2/q5dX3Oc0bNjQsH3r1i02bdrkVC5bFni0ulhftxPOXzYVyJQNStSE\nnUsM2ZnJQhGKA3CUg9zhdprVPd2w2aBGFGzaktE1AaBNYVh/Ec7eyeiaKO4XChUqxNmzZ0lISMjo\nqigUlijh+F6m3evQ8gV9+8p5uHNL386eR+TlyAuXz4llSBhcOgextyFHPggJvfv1llg7AvJnE+t5\nXoEL1zO0Oor7hLx58zJ58mSKFy9uyI+Li+Ptt992CMili8J34yA0GIoXFGWavwFLN5hPGAldpsPG\nH+Cb/gC0pRNFKU4EOQD4H+M5xYn0/Flpw4TRUKggHP8X+r2XoVWZUh3eKA0nbkHRnzK0Kilm4wqx\nrNXIh4P8/KHL7/r2Z83TtE7pRUGK0pYX+Zej/MTXGVKHmTNnUqBAAW7evEm7du0ypA6pIl8ueEf6\nLp+6AJMy5l4+qOzSlhUzsA7KrOJeJXcBo2B85xZggysXoMBDIoVmEsuQMIjIBTGa1itnXqFBHv1K\nhlRd5sxVsTxyXmmNFd7ToEEDihcvzu3bt+nSpQvr168HICgoiOnTpzvKDX8V9h3VBWOXvDVPLHOI\ngnnITwABXONqelQ/fSmk/dhCBTK0GkUzC60xQJFMGVqVVFG2lEgjve1nlGkJrSdDrpJ6uk/owQDy\nkJ9CFCOMzHf12n/88Qdbt26lUqVK5MyZk2LFit3V66cJJYvA5L5wOwau3oBCeaF2RfhuPFSIzOja\n3ff0BJKA8lral4F1UcLx/cC3U2FoJxjZFSb3gz2b9X1njsOgDjDsJRjRFT7WWvigECE0fzI0I2rs\noMkoKNsb6g2B6/fBiHV68P1HH/H9Rx9ldDXuG3r27Mk777wDwHPPPcfOnTuZNGmSZdlnBorUvr+H\nk87pI5Zf9wXgPGf4lpn8xRI3B92j9BmU0TUA4Hg0bNHMV7o7W7o8uOxbCL+8ldG1SBVxxBJP3F29\nZpcuXWjXrh3R0dF39bppRt4c8G5Xsf7bahg3G/45ou8/dylDqvUgsQL4LaMroaGE43udmNtwZDdc\nOA1XL0JCPMRIphXxcUJAvnhGmFhsXgG7NkBcjDg2d6EMqzrAtVuw7zScu5ah1cgwqpUrR7GCBbl8\n7T96A1JArly5HOunTp0C4NixY0RFRTFnzhyioqKcjjl+xsNJt/wCbxaHq54K3gecPAXtX4QOnTO6\nJrT+C2xz4H8HM7omKSe8IMz9HgaP8uGgpERY93G61clbdt+GR1p4X34Sw5nIMGYwPn2F49DCUH87\nlNJv6r///suxY8dISkpKv+umJ9XLC/utE2dh8x6IT4Ad+2Gn9vBnzSw0yA8wNmz057N0O/8+4Ml0\nO7tvKJvje5XL5+HvtfD1RLh01rdjZ40Bm58QlvPfh0NXDwilihVj8iCh5Zv5ww8ZXJsHg8mTJ2d0\nFRQPIN3fzuga+M57KRiMuoCP35KUElELArLenWvdLV5o6Zy3YBUsXgc5I2BwxpsxpjddGZnRVbhr\nKOH4XiUxAT7qm7Jjr17U16+cN+4Lzw7lqhvz9myCmy40m7XCxbJYCDTJBp9qjevGG15VpXn58o71\nFfv3E2fyTq70vFhe3AdntuvbdnbOxbI8QHIS7Jon7/UHWmnre4AjCMslu1NXLHgYRs9BCTKTy5B3\nArN3lyA8UybqVq3K9eho1u/YQfP69Q37R76lD71Wr1iRnNmEZ+LiNWsM5dpK6wOAscBywPwfKQjU\n0taXAjeBZuCwHDwObHXz28iHfitk1loXjwTKmPIWuDu/BYE2G41z5jTk7YmO5uQdY2iD0NBQGjRo\nAECBArotbfPmwtEpOjqaNab75jONEON2Eje5zm2iLe0vS1KOg/yTums6KIh+N/8CvDPAj6AO19gA\nJBvyQyhMDP+6PzgAaKhd6k/XxUIQt8bOn0A24LRXNYRw7RwXPBWsh3jW6gLrvDy5wpLwCKjgPIBy\n71B6TEbXIP3Ysd+4HRwExQsJAVk2s3jAqMcTZCdPRlfjrqGE4/8aI76ECKOwwtWL8PYTzmXfLACV\nTULDK/nEslgIzHP9OWxcpgw9GzakoCYQAmyaNIkrmnD89mGRl10T1tZ9CE9/CXnKG8+TOQ+smwi1\nekLtXnp5AJIhJAI2zwB4BXgOsJuRjAbOAV8DWbS8JOAUIIXHk2jJOEIIJ4AQQ35R6rKKCY7tIvnz\nM3nQIAIDA8mTIwf7jx6lYc2atG7c2OX9ePnppx3rsnA8EzDf+RnAGaCSlPctwnPXLrbXBuoAXwCB\nWl60lmepGxoN5EZIMmZyAT8bs5YAOYAIU9E3Ade/0pkfq1Ujf4jxfl6Nj6eJKRxb1qxZGTnSWSth\nz0tISKBPnz6sW5cCqepLxA3Lh/idL+q7QgglSPp/P0RJWiN6YFmJ4DrX+IqPuYSpk8l0aX0oMFza\nHgxckbZnAPnR72ZXYJFWMZ1IxuBHMAAH6E1JxhNMfvLQlsss4xJLyElTctAUfzKRwA0OYtGBnq8t\n/bTfnAS8BjzjXPQtoIVWOztngWDgEefiBooBk4Ag7VLNrApNAqpp63mA89qyHbpLukemW+SN0E4G\nrHZx2MPa8kmgj2lfS0TvUtvdGfgFmO2uGt1NPYxQz5rRRaZtSwuI7ohOzHREvwnge2n/Rzh1Jn7c\nAnklf8zIcjB5PrQwtZ9mwihBIV7nAL1dlJgHDAKOSdsAm7SKeCBXcyjezzm/WC845vr41q1b89JL\nL/HTTz/x1Vdfub+GP0ZPrViggueqmckRGMjMSqKVnXr8OCsuubAZnmpyaKhdUSSAT3+EW3fgzQ5i\nO39u3yviAns/3t5xbQAcBWOXeDDcLWVuNRp6LpQKxgGeXEfuJrbk5ORkz8XS+KI2292+5INFz/f1\n+McnDgpnPVdEVoR3PzHm7d4EFWoa816qbdyunFkIx1fiYf8dOB4DWQOgZXa9zMBjcN7abm34E0/Q\nqEwZQgIDHXmPTZrElVu3GHgJrh6FAtWdj7vwD+QsLSIl2fmpMzw9S6yf2wV5pfgu//wA37YLQOhb\nn7KoyS6cA8JcAvTfH0AwD/M2
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 6
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"So, the `label` function assigns a different color to each connected component.\n",
|
||
|
|
"\n",
|
||
|
|
"A connected component is defined on the *pixel adjacency graph*.\n",
|
||
|
|
"\n",
|
||
|
|
"We define two ON pixels as neighbors if they are next to each other horizontally or vertically.\n",
|
||
|
|
"\n",
|
||
|
|
"**Definition:** Two pixels are in the same connected component if they are (immediate) neighbors \n",
|
||
|
|
"or if the first pixel is in the same connected component as a neighbor of the second pixel.\n",
|
||
|
|
"\n",
|
||
|
|
"This definition is consistent with the definition of the connected coponents of a graph.\n",
|
||
|
|
"\n",
|
||
|
|
"There are various efficient algorithms for connected component labeling,\n",
|
||
|
|
"including parallel algorithms.\n",
|
||
|
|
"Asymptotically, these algorithms run in (nearly) linear time."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "heading",
|
||
|
|
"level": 1,
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Algorithms for Connected Component Labeling"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Let's look at the connected component labeling algorithm.\n",
|
||
|
|
"\n",
|
||
|
|
"We're going to present a serial algorithm (i.e., without Python array optimizations)."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"We keep track of the labels in a new array `a`.\n",
|
||
|
|
"\n"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"a = -array(image,'i')"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [],
|
||
|
|
"prompt_number": 7
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"For implementing connected component labeling, we need a data structure that keeps track of of sets of pixels.\n",
|
||
|
|
"\n",
|
||
|
|
"The operations are:\n",
|
||
|
|
"\n",
|
||
|
|
"- make a new set and return an integer identifier\n",
|
||
|
|
"- find a canonical representative for a given set\n",
|
||
|
|
"- make a union of two sets\n",
|
||
|
|
"\n",
|
||
|
|
"When we are all done, we can map each element of a set to a single representative."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"class UnionFind:\n",
|
||
|
|
" def __init__(self):\n",
|
||
|
|
" self.parents = []\n",
|
||
|
|
" def set(self):\n",
|
||
|
|
" i = len(self.parents)\n",
|
||
|
|
" self.parents.append(i)\n",
|
||
|
|
" return i\n",
|
||
|
|
" def find(self,i):\n",
|
||
|
|
" if self.parents[i]==i: return i\n",
|
||
|
|
" self.parents[i] = self.find(self.parents[i])\n",
|
||
|
|
" return self.parents[i]\n",
|
||
|
|
" def union(self,i,j):\n",
|
||
|
|
" i = self.find(i)\n",
|
||
|
|
" j = self.find(j)\n",
|
||
|
|
" if rand()<0.5: self.parents[i] = j\n",
|
||
|
|
" else: self.parents[j] = i"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [],
|
||
|
|
"prompt_number": 8
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"This is the meat of the algorithm. There are four cases:\n",
|
||
|
|
"\n",
|
||
|
|
"- off pixels are just skipped\n",
|
||
|
|
"- if a pixel is on and the pixel to the left is on, it assumes the label of the pixel to the left\n",
|
||
|
|
"- if a pixel is on and the pixel above is on, it assumes the label of the pixel above\n",
|
||
|
|
"- if there is a conflict, we make a union of the two sets"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"table = UnionFind()\n",
|
||
|
|
"\n",
|
||
|
|
"for i in range(1,a.shape[0]):\n",
|
||
|
|
" for j in range(1,a.shape[1]):\n",
|
||
|
|
" if a[i,j]==0: \n",
|
||
|
|
" pass\n",
|
||
|
|
" elif a[i,j-1]>0: \n",
|
||
|
|
" # pixel to the left is on\n",
|
||
|
|
" a[i,j] = a[i,j-1]\n",
|
||
|
|
" # both left and above are on\n",
|
||
|
|
" if a[i-1,j]>0: table.union(a[i-1,j],a[i,j])\n",
|
||
|
|
" elif a[i-1,j]>0:\n",
|
||
|
|
" # only pixel above is on\n",
|
||
|
|
" a[i,j] = a[i-1,j]\n",
|
||
|
|
" else:\n",
|
||
|
|
" # ths pixel is on, but neither left or above are\n",
|
||
|
|
" a[i,j] = table.set()\n",
|
||
|
|
" "
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [],
|
||
|
|
"prompt_number": 9
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"After labeling, we map each pixel to its canonical representative.\n",
|
||
|
|
"\n"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"for i in range(1,a.shape[0]):\n",
|
||
|
|
" for j in range(1,a.shape[1]):\n",
|
||
|
|
" a[i,j] = table.find(a[i,j])"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [],
|
||
|
|
"prompt_number": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"The canonical representatives chosen by `UnionFind` aren't contiguous, so we need to remap the labels. Note that this ends up remapping 0 to 0.\n"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"remap = {v:i for i,v in enumerate(unique(a))}\n",
|
||
|
|
"assert remap[0]==0\n",
|
||
|
|
"\n",
|
||
|
|
"for i in range(1,a.shape[0]):\n",
|
||
|
|
" for j in range(1,a.shape[1]):\n",
|
||
|
|
" a[i,j] = remap[a[i,j]]"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [],
|
||
|
|
"prompt_number": 11
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Let's look at the output.\n",
|
||
|
|
"\n"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"figsize(8,8)\n",
|
||
|
|
"imshow(a)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "pyout",
|
||
|
|
"prompt_number": 12,
|
||
|
|
"text": [
|
||
|
|
"<matplotlib.image.AxesImage at 0x3bf6510>"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAAAeoAAAGRCAYAAACqiPA4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCep0jGrflRZRVCe+0TKrk6av
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 12
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Or the output in color, as before.\n",
|
||
|
|
"\n"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"figsize(8,8)\n",
|
||
|
|
"imshow(recolor(a),cmap=cm.spectral)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "pyout",
|
||
|
|
"prompt_number": 13,
|
||
|
|
"text": [
|
||
|
|
"<matplotlib.image.AxesImage at 0x3c0d610>"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAAAeoAAAGRCAYAAACqiPA4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngiI1xK7p1w8hJtrBZQCRcs8U
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 13
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Operations over Labeled Regions\n",
|
||
|
|
"===============================\n",
|
||
|
|
"\n",
|
||
|
|
"There is a variety of image processing operations that can be carried out\n",
|
||
|
|
"over labeled regions.\n",
|
||
|
|
"\n",
|
||
|
|
"For example, the `measurements.sum` function (not to be confused with the `pylab.sum`\n",
|
||
|
|
"function), sums all the values of the first arguments over each of the regions\n",
|
||
|
|
"defined by the labels of the second argument, for the labels listed in the third\n",
|
||
|
|
"argument.\n",
|
||
|
|
"\n",
|
||
|
|
"Applied to the image above, this gives us the distibution of sizes of\n",
|
||
|
|
"each of the connected component (since the image has just values of 0 and 1)."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"sizes = measurements.sum(image,labels,range(n+1))\n",
|
||
|
|
"_=hist(sizes,bins=50)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAAAeoAAAHcCAYAAAAZTumKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nGtr1fezYsVx33XUZHBxMkvzo
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 14
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"This combines in interesting ways with subscripting.\n",
|
||
|
|
"\n",
|
||
|
|
"We have a 1D array of sizes and a 2D array of labels.\n",
|
||
|
|
"Subscripting with arrays returns, for each element of the subscript,\n",
|
||
|
|
"the value of the array being subscripted.\n",
|
||
|
|
"\n",
|
||
|
|
"So, here is a simple way of removing small or large connected\n",
|
||
|
|
"components from an image."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"lsizes = sizes[labels]\n",
|
||
|
|
"imrow(lsizes,lsizes>80,lsizes>300)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAABWcAAAFlCAYAAACHqVoEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsnXlcVcX7xz/3XvZFkB1EUREEJBcUxQVcUNw1NVBTQcQN\nNauv2aKBqbllZYtWGoamqWlpZi65pLlhLqm5ZW5lJS6AC7Jzmd8f9zfTOXdj8bKoz/v1mhfcOefM\nzJmzzJzPPPOMgjHGQBAEQRAEQRAEQRAEQRAEQVQpyuouAEEQBEEQBEEQBEEQBEEQxNMIibMEQRAE\nQRAEQRAEQRAEQRDVAImzBEEQBEEQBEEQBEEQBEEQ1QCJswRBEARBEARBEARBEARBENUAibMEQRAE\nQRAEQRAEQRAEQRDVAImzBEEQBEEQBEEQBEEQBEEQ1UCliLM7duxAQEAA/Pz8sGDBgsrIgiAIgiCe\nSqiNJQiCIAjTQ+0rQRAEUV0oGGPMlAmq1Wo0btwYu3fvRp06dRAaGoq1a9ciMDDQlNkQBEEQxFMH\ntbEEQRAEYXqofSUIgiCqE5Nbzh49ehSNGjVC/fr1YW5ujiFDhmDz5s2mzoYgCIIgnjqojSUIgiAI\n00PtK0EQBFGdmJk6wX///Rd169YVv729vfHLL7/I9lEoFKbOliAIgnjCMPHEjicCamMJgiAIU0Bt\nrBxqXwmCIAhTUNH21eSWs9RoEQRBEETlQG0sQRAEQZgeal8JgiCI6sTk4mydOnXw999/i99///03\nvL29TZ0NQRAEQTx1UBtLEARBEKaH2leCIAiiOjG5ONuqVStcunQJf/75JwoLC/H111+jX79+ps6G\nIAiCIJ46qI0lCIIgCNND7StBEARRnZjc56yZmRkWL16M7t27Q61WIyEhgVa5JAiCIAgTQG0sQRAE\nQZgeal8JgiCI6kTBqsEbPPn0IQiCIEqDFiupGNTGEgRBEKVBbWz5ofaVIAiCKI0asyAYQRAEQRAE\nQRAEQRAEQRAEUTokzpaBli1bomXLluK3l5cXzMzMxDb+lzGGJUuWmCRPpZIujSlxdnYWQYqVlRVq\n1aoli5OOiisUCnh4eFRJGSvjmtN9ZDrc3d2ruwgE8URSHW0sYVr4NZReRwCoW7euTpwUhUJhdDvx\n9BASElLdRSCIJw5qXwmCIB4fTO5z9knj3XffxZQpU0rdb9q0aSbL09LSEgqFAvn5+SZL82mmc+fO\nqF+/Pu7duwdHR0fcvn0bW7duRY8ePcQ+O3bsEP9bWFigsLAQgYGB6N+/PwBg3rx5lV5OCwsLFBUV\nQa1WmyQ9S0tLKJVK5OXlmSS9smBlZYWSkhIUFhbqbPPw8EBCQgIsLCywYMEC5ObmVnpZ1Go1ioqK\nHimdvn37wsPDA7Vr18Y777xjotIZR6VS4ZNPPgEA3LhxAzNnzqySfCuLkSNHomHDhjh//jzWrVtX\n3cUhahDV0cYSpmX9+vWIjo7G+fPnERQUhLS0NLRr1042pUvfVOChQ4dizZo1BrcT5aN58+Y4cOAA\n7Ozs4OrqioyMjOouUplYu3YtWrZsCT8/vyq7D8zMzETf4OzZs3jmmWeqJN/KYvfu3YiMjMTGjRsx\naNCg6i4OUUOg9pUgCOLxgszqSqFp06YANB8OCoUCOTk5st8NGzYEYFrxrqCgoEYKs5aWlli4cGF1\nF6Nc2Nraon79+khNTcWmTZuQkZEh6nbHjh3466+/dI4pKSkBYwznz5+vMjFOmq+pKCwsNJnQW1ZK\nSkoMbrt58yZ+/fVXk+anbQldnvKUhouLCwBgy5YtOHHiRIXTqQiffPIJFi9ejDfffBNeXl6ws7OD\ntbV1lZZBH66urhU6bsWKFQA0VnQEIaU62tiaip2d3WPnA9Lb2xvR0dFQKBRo0qQJTpw4gTt37gDQ\nXMMPP/zQ4LFr166FpaVlVRX1iefUqVNISUkxaZqVuVI9T3vo0KH4+OOPKy0ffRQVFaFPnz5o1KgR\ngoOD4ezsDAcHhyotgz4qWt9du3YFALRt29aUxSEec6h9JQiCeLwgcbYURo4cCS8vL4Pbr127Bj8/\nP9ja2lZhqaqHt99+G9988011F+OR2LJlC/bs2WN0n0e1tKwohYWFjyQmasMY02vBWpkUFhYazXP7\n9u0my8vV1RXBwcEGt+fn51dYnLa3t682S5ru3bsDAM6cOYM7d+7g1VdfRVJSkhCLq5NmzZpVdxGI\nJwxqY/8jOzsbY8eOre5iPBKtWrUSM06Iqufll182WVoxMTHYvHmzydKT0rJly0pLuzRee+01AMDW\nrVtx5coVeHl54fTp00Koqk6qq06IJxNqXwmCIB4vSJwthRs3biA9Pd3oPpcvXxajkYBG3GOMgTEG\nf39/EW9nZ4f4+HgRunTpopOWpaUlQkJC0Lp1axHXs2dPbNmyBQsXLkRKSgq2bNmCLVu2yLZt2bIF\nEydOBAC0adMGc+bMgZ2dHYYNG4Y5c+ZgyJAhOnmFh4eLskjL2bZtWyxcuBCenp5YuHChCADw3HPP\nyaxn9+/fjwsXLuDChQv49NNPRfzkyZNx4cIFAMCBAwdw4cIFxMbGyvL39fVFdHS0CFLxKSAgQFZX\nxkhISBDB3NxcxMfHxyMmJkb8X1o6AODo6Ih69eoJX61SsTQkJASRkZEyv2j29vaoU6cObG1tUadO\nHdSpU0dsCwoKwrhx49CzZ09ZHk2bNkWDBg3g7++P6OhoKJVKuLu7o0WLFqhdu7bYz87ODoGBgRgz\nZoyOCOnn54dhw4Zh2LBhBqewubm5iVFzTqNGjeDn5wc/Pz+0adPGYD0EBgaKwPHx8RFBO4775bW0\ntNRrXRkaGiqCPho3bozGjRvL4po0aSLSbdq0qc65BAUFAdBcA304OTnJfqtUKjg6OgIAgoODZech\npXbt2mjVqhUAjWsEbczNzdG8eXO9xwYEBCAgIAC+vr56t0tp3LixTtnbtm2Lhg0boqSkBB07dkTr\n1q0xbNgwODo64u7du6WmyeHXVqVSibhHtVo1dO1Kg3z1EsYwZRvboEEDEc8Yw7fffquTVsOGDZGZ\nmQnGmJhCnZiYCMYYDh8+jKtXr4rjpdsYY1i6dKkszsPDA5s2bQJjTK+7jtTUVHHsmDFjRDw/vmnT\nprLyAsCyZctk1rM3btwQ23m7DwCzZ88W+6Wnp4MxhhdffFGWv7TsjDGEh4cb3GYM6X52dnay+L//\n/lu2T2kMGjQIy5cvh42NDQCguLhYVqb3338fiYmJIi4kJASJiYlo0qQJEhMTZdv4OWzdulXnvPv0\n6YPBgwfj4MGDsLCwQGhoKF5++WUEBASI/erVqyfSMFZ3t2/f1nsuiYmJuHTpkixu3LhxopzGrFj5\nPtrnYyju2WefBaBp+6XbjR2rb7uUUaNGife6vu1ff/01AMPvfu1+pZWVlWjHExMTdfo+nMjISBw/\nfhyA5pnVxsbGxuB5jB8/HomJiRgxYoTe7VISExN1BNe4uDiEhYWhuLgYiYmJeP755/Hpp5+iTp06\n4l4uC7x80j5n7969y3y8PtLS0ip03OPukoGoPEzZvlaU8ravVcmjtK8EQRCVgYJVwxy6x9m32MOH\nD2Fra6v3HPRV5YgRI7B69WoAGoFw48aNyM7ORmxsLBQKBVJTU2X7K5VKIQwdPXoUCQkJolOujVqt\nhkqlwtatW0WnsF+/fkhMTBQi4cGDB9GhQwcAQFZWFt577z0AmtFUnn/z5s3RokULpKWl4eLFi3qn\n8k+bNg1z584V/xcVFaF9+/ZISUlBixYtkJ+fjwsXLiAwMBDm5ub47bff9Jb5vffeEx8s0dHROHDg\nAMzMzMRUrA0bNqBFixZo3rw5Dh48iKtXryI2NhaMMTE1WkpCQgIyMjLwww8/YOTIkQCA5cuXi7q0\ns7PDoEGD8OWXX4o6kxIYGAgfHx/s2LEDtra28PPzAwCcO3dOWNC+8cYbKC4uFg70AeD06dPYtm0b\nOnfurFOmvXv3YuzYsbh8+TL2
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 15
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Finding Objects\n",
|
||
|
|
"===============\n",
|
||
|
|
"\n",
|
||
|
|
"Although labeling is useful by itself, often, we want to _extract_ images corresponding\n",
|
||
|
|
"to the labels.\n",
|
||
|
|
"\n",
|
||
|
|
"For doing this, we use `measurements.find_objects`.\n",
|
||
|
|
"It returns a list of pairs of slices.\n",
|
||
|
|
"Each `slice(a,b,c)` is a data structure representing\n",
|
||
|
|
"array subscript notation like `a:b:c`.\n",
|
||
|
|
"So, the first element of the result represents the\n",
|
||
|
|
"subscript `[15:17,338:341]`, but without any array\n",
|
||
|
|
"to be subscripted.\n",
|
||
|
|
"\n",
|
||
|
|
"These two slices together define the _bounding box_ of a connected component."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"# find objects finds connected components\n",
|
||
|
|
"objects = measurements.find_objects(labels)\n",
|
||
|
|
"objects[:3]"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "pyout",
|
||
|
|
"prompt_number": 16,
|
||
|
|
"text": [
|
||
|
|
"[(slice(21L, 51L, None), slice(111L, 155L, None)),\n",
|
||
|
|
" (slice(22L, 26L, None), slice(185L, 190L, None)),\n",
|
||
|
|
" (slice(22L, 51L, None), slice(394L, 411L, None))]"
|
||
|
|
]
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 16
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"(slice subscripting)\n",
|
||
|
|
"\n",
|
||
|
|
"We can directly subscript with tuples of slices.\n",
|
||
|
|
"\n",
|
||
|
|
"When we do this for the result of `find_objects`, we get pictures of all\n",
|
||
|
|
"the connected components.\n"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"subimages = [image[s] for s in objects[10:30]]\n",
|
||
|
|
"imrow(*subimages,s=1)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAABHUAAABVCAYAAADHcdgIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nPPLII4iPj5/l7qEJTS47N2/e
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 17
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"(subscripting with label selection)\n",
|
||
|
|
"\n",
|
||
|
|
"But the above is not quite right: if the bounding boxes of two connected components overlap,\n",
|
||
|
|
"we possibly get pixels from both connected components within each bounding box.\n",
|
||
|
|
"We actually need to _mask_ the image with a mask computed from the labels.\n",
|
||
|
|
"\n"
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"images = [image[s] * (labels[s]==i+1) for i,s in list(enumerate(objects))[10:30]]\n",
|
||
|
|
"imrow(*images,s=1)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAABHUAAABVCAYAAADHcdgIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8eijjyI+Pn6Wu4cmNLns3Lx5
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 18
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"(connected component measurements)\n",
|
||
|
|
"\n",
|
||
|
|
"Now that we have the isolated images, we can compute properties of each object in isolation.\n",
|
||
|
|
"\n",
|
||
|
|
"Common measurements on isolated objects are the _aspect ratio_ and the _bounding box area_.\n",
|
||
|
|
"\n",
|
||
|
|
"As this example shows, these simple measurements already characterize objects quite well.\n",
|
||
|
|
"\n",
|
||
|
|
"The example also shows that dividing an image into components by bounding boxes does not\n",
|
||
|
|
"always work, since the characters are not actually separated in the images of `wt`, `as`,\n",
|
||
|
|
"`rm`, and `er`. \n",
|
||
|
|
"\n",
|
||
|
|
"When we are interested in indivual characters, these images are _segmentation errors_."
|
||
|
|
]
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "code",
|
||
|
|
"collapsed": false,
|
||
|
|
"input": [
|
||
|
|
"# shape descriptors\n",
|
||
|
|
"def aspect(s):\n",
|
||
|
|
" return (s[0].stop-s[0].start)*1.0/(s[1].stop-s[1].start)\n",
|
||
|
|
"def bbarea(s):\n",
|
||
|
|
" return (s[0].stop-s[0].start)*(s[1].stop-s[1].start)\n",
|
||
|
|
"tall = [image[s] for s in objects if aspect(s)>2.0]\n",
|
||
|
|
"imrow(*tall[:20],s=1)\n",
|
||
|
|
"wide = [image[s] for s in objects if aspect(s)<0.7 and bbarea(s)>100]\n",
|
||
|
|
"imrow(*wide[:20],s=1)"
|
||
|
|
],
|
||
|
|
"language": "python",
|
||
|
|
"metadata": {},
|
||
|
|
"outputs": [
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAABGQAAABVCAYAAAAPHTLeAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSW793ZIrF3Rl0kdoaCg2bdoE
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"output_type": "display_data",
|
||
|
|
"png": "iVBORw0KGgoAAAANSUhEUgAABH0AAAA+CAYAAAC2lWdlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYwd+/fVX3gIkierqapw5cwZp
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"prompt_number": 19
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"cell_type": "markdown",
|
||
|
|
"metadata": {},
|
||
|
|
"source": [
|
||
|
|
"Of course, the next step after this kind of segmentation is to try to recognize\n",
|
||
|
|
"the characters. We will be looking at techniques for doing this later in the course."
|
||
|
|
]
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"metadata": {}
|
||
|
|
}
|
||
|
|
]
|
||
|
|
}
|