machinelearning_notebook/3_kmeans/2-kmeans-color-vq.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 用Ｋ-means进行颜色量化\n",
    "\n",
    "本示例对给定的图像进行**像素矢量量化(VQ)**，将显示图像所需的颜色从16777216（$2^{24}$)种减少到64种，同时保持整体外观质量。\n",
    "\n",
    "\n",
    "在本例中，像素在3D空间中表示，使用K-means找到64个颜色簇。在图像处理文献中，由K-means(聚类中心)得到的聚类中心集合称为调色板，例如，GIF文件格式就使用了类似这样的调色板。使用单个字节，最多可以寻址256种颜色，而RGB编码需要每个像素3个字节，因此采用本方法可以简单的将原始图像压缩到1/3。\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.metrics import pairwise_distances_argmin\n",
    "from sklearn.datasets import load_sample_image\n",
    "from sklearn.utils import shuffle\n",
    "from time import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7fdcf8ebb1d0>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {