median_doesnt_work_wo_normalization.ipynb

{
 "metadata": {
  "name": "median_doesnt_work_wo_normalization"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import pyfits"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "fils = ['flats-271R_flt.fit', 'flats-272R_flt.fit', 'flats-273R_flt.fit']"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 2
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "In the description that follows these images will be referred to by their 'python' numbers: 0, 1, 2"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "all_hdus = []\n",
      "for fil in fils:\n",
      "    all_hdus.append(pyfits.open(fil))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "shape = list(all_hdus[0][0].data.shape)\n",
      "shape.append(3)\n",
      "med = zeros(shape)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "med.shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "pyout",
       "prompt_number": 5,
       "text": [
        "(2048, 3085, 3)"
       ]
      }
     ],
     "prompt_number": 5
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "med_values = zeros(3)\n",
      "for idx, hdus in enumerate(all_hdus):\n",
      "    med[:,:,idx] = hdus[0].data\n",
      "    med_values[idx] = median(med[:, :, idx])"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 6
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "med_values"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "pyout",
       "prompt_number": 7,
       "text": [
        "array([ 22462.,  22891.,  23172.])"
       ]
      }
     ],
     "prompt_number": 7
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "npmed = median(med, axis=2)\n",
      "new_med = med.copy()\n",
      "ref_value = med_values[2]\n",
      "for idx in range(0,3):\n",
      "    new_med[:,:,idx] = new_med[:,:,idx]/med_values[idx]*ref_value\n",
      "npnew_med = median(new_med, axis=2)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 8
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "crapper = pyfits.PrimaryHDU(data=npmed)\n",
      "crapper2 = pyfits.PrimaryHDU(data=npnew_med)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 9
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "crapper.writeto('python-median.fit', clobber=True)\n",
      "crapper2.writeto('python-median-norm.fit', clobber=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Overwriting existing file 'python-median.fit'.\n",
        "Overwriting existing file 'python-median-norm.fit'.\n"
       ]
      }
     ],
     "prompt_number": 10
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "In principle, a simple median combine should eliminate artifacts like cosmic rays. In practice that will work, I think, only if the background level in the images is about the same.\n",
      "\n",
      "If it isn't, the defect will tend to bias the median a bit higher in the pixels where the defect shows up than in the pixels where the edefect is not present."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Consider the pixel values below, at the position of one of the fake cosmic ray. The pixels values are frome images 0, 1, 2, andthe ray is in the first image. In this case the median will be from image 2 (python numbering for the images, this is 3_flt)."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "med[819,1480,:]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "pyout",
       "prompt_number": 11,
       "text": [
        "array([ 24495.76367188,  23260.        ,  23492.        ])"
       ]
      }
     ],
     "prompt_number": 11
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "In an adjacent pixel without a cosmic ray the median value is from image 1:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "med[820,1480,:]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "pyout",
       "prompt_number": 12,
       "text": [
        "array([ 22705.,  23027.,  23544.])"
       ]
      }
     ],
     "prompt_number": 12
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "This will be the case for most pixels because the median of all of the pixels in image 1 is higher than the median for image 0 and lower than the median for image 2:"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "med_values"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "pyout",
       "prompt_number": 13,
       "text": [
        "array([ 22462.,  22891.,  23172.])"
       ]
      }
     ],
     "prompt_number": 13
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "The net result, then, is for the result of the median to be slightly higher where the cosmic ray is because those pixels always have, as their highest value, the contribution from the cosmic ray. If the remaining images all have about the same pixel value that doesn't matter. In this case, though, choosing a pixel from image 2 means choosing a pixel that is roughly 280 ADU higher than is typical for surrounding pixels (which will end up being taken from image 1).\n",
      "\n",
      "The cosmic ray is *reduced*, but not eliminated.\n",
      "\n",
      "Normalizing the images before combining does fix the problem. For this toy example I've just normalized by scaling the raw flats to a common value for the image median. That isn't really the right thing to do; they should be dark/bias subtracted first, then normalized."
     ]
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "I fear the problem is due to variation in the bias level rather than the lighting. If so, we should see it in the overscan, which is columns 3072-3084 (inclusive); note that column is the second index.\n",
      "\n",
      "The plot below shows the overscan region for each flat, averaged over all of the rows for that flat. Only the middle part of the overscan region is shown because it rises sharply on the ends."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "for idx in range(med.shape[2]):\n",
      "    oscan = med[:,3073:,idx].mean(axis=0)\n",
      "    plot(oscan[1:8], label=str(idx))\n",
      "    \n",
      "legend()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "[  5870.03613281   3474.41699219   3151.765625     3100.84423828\n",
        "   3094.12402344   3102.70898438   3159.34082031   3544.47998047\n",
        "   6109.23730469  22612.11181641  10171.75439453   3129.73339844]\n",
        "[  5933.59228516   3483.36621094   3153.48779297   3103.05908203\n",
        "   3095.10253906   3102.70800781   3161.75585938   3555.44287109\n",
        "   6177.64501953  23196.47802734  10173.59375      3128.86621094]"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "[  5973.98779297   3490.23535156   3153.94091797   3101.96386719\n",
        "   3095.42480469   3102.71044922   3162.46777344   3563.66943359\n",
        "   6227.15429688  23593.73291016  10175.88378906   3128.64550781]\n"
       ]
      },
      {
       "output_type": "pyout",
       "prompt_number": 14,
       "text": [
        "<matplotlib.legend.Legend at 0x34c94f0>"
       ]
      },
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAXsAAAD9CAYAAABdoNd6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n"
      }
     ],
     "prompt_number": 14
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Clearly the bias level (to the extent it is accruately represented by the overscan) is **not** varying by a few hundred ADU from frame to frame.\n",
      "\n",
      "Good! Except that it means the floodlight isn't so steady..."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 14
    }
   ],
   "metadata": {}
  }
 ]
}