mypapers/drafts/geg2210/assignment-8.html

   1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
   2 <html>
   3   <head>
   4     <title></title>
   5   </head>
   6
   7   <body>
   8
   9     <p><a href="http://www.geo.uio.no/geogr/geomatikk/oppgaver/bildeforbedring_eng.html">Assigment 8</a>
  10       in <a href="http://www.uio.no/studier/emner/matnat/geofag/GEG2210/index-eng.html">GEG2210</a>
  11       - Data Collection - Land Surveying, Remote Sensing and Digital
  12         Photogrammetry</p>
  13
  14     <h1>Image enhancement, filtering and sharpening</h1>
  15 <img width="40%" src="jotunheimen-std-ir-eq.jpeg">
  16
  17     <p>By Petter Reinholdtsen and Shanette Dallyn, 2005-05-01.</p>
  18
  19 <p>This exercise was performed by logging into jern.uio.no using ssh
  20 and running ERDAS Imagine.  Started by using 'imagine' on the command
  21 line.  The images were loaded from /mn/geofag/gggruppe-data/geomatikk/
  22
  23 <p>We tried to use svalbard/tm87.img, but it only have 5 bands.  We
  24 decided to switch, and next tried jotunheimen/tm.img, which had 7
  25 bands.
  26
  27 <p>The pixel values in a given band is only a using a given range of
  28 values.  This is because sensor data in a single image rarely extend
  29 over the entire range of possible values.
  30
  31 <p>The peak values of the histograms represent the the spectral
  32 sensitivity values that occure the most often with in the image band
  33 being analysed.
  34
  35 <h2>Evaluation of the different bands</h2>
  36
  37 <p><img align="right" width="40%" src="jotunheimen-truecolor.jpeg">
  38 This image show the "true colour" version, with the blue range
  39 assigned to the blue colour, green range to green colour and red range
  40 to red colour.</p>
  41
  42 <h3>band 1, blue (0.45-0.52 micrometer - um)</h3>
  43
  44   Visible light, and will display a broad range of values both over
  45   land and water.  Reflected from ice, as those are visible white and
  46   reflect all visible light waves.  Histogram show most values between
  47   30 and 136.  Mean values of 66.0668.  There are one wide peak with
  48   center around 50.  There are two peaks at 0 and 255.
  49
  50 <h3>band 2, green (0.52-0.60 um)</h3>
  51
  52   Visible light, and will display a broad range of values both over
  53   land and water.  Reflected from ice, as those are visible white and
  54   reflect all visible light waves.  Histogram show most values from 8
  55   to 120.  The mean value is 30.9774.  There are two main peaks at 20
  56   and 27.  There is also a pie at 0.
  57
  58 <h3>band 3, red (0.60-0.69 um)</h3>
  59
  60   Visible light, and will display a broad range of values both over
  61   land and water.  Reflected from ice, as those are visible white and
  62   reflect all visible light waves.  Histogram show most values from 33
  63   t 135, with one wide peak around 52.  There are also seem to be two
  64   peaks at 0 and 255.  The mean value is 34.3403.
  65
  66 <h3>band 4, near-infraread (0.76-0.90 um)</h3>
  67 <img align="right" width="20%" src="jotunheimen-band4-hist.jpeg">
  68
  69   Water acts as an absorbing body so in the near infrared spectrum,
  70   water features will appear dark or black meaning that all near
  71   infrared bands are absorbed. On the other hand, land features
  72   including ice, act as reflector bodies in this band.  The histogram
  73   show most values between 7 and 110.  The mean is 40.1144.  There are
  74   two peaks at 7 and 40.
  75
  76 <h3>band 5, mid-infrared (1.55-1.75 um)</h3>
  77
  78   The ice, glaciers and water do not reflect any mid-infrared light.
  79   The histogram show most values between 1 and 178.  The mean is
  80   49.8098 and there are two peaks at 6 and 78, in addition to two
  81   peaks at 0 and 255.
  82
  83 <h3>band 6, thermal infrared (10.4-12.5 um)</h3>
  84
  85   Display the temperature on earth.  We can for example see that the
  86   ice is colder than the surrounding areas.  The histogram show most
  87   values between 36 to 122.  The mean is 102.734.  There are one wide
  88   peak around 53, in addition to two peaks at 0 and 255.
  89
  90 <h3>band 7, mid-infrared (2.08-2.35 um)</h3>
  91
  92   The ice, glaciers and water do not reflect any mid-infrared
  93   frequencies.  The histogram show most values between 77 and 150.
  94   The mean is 24.04, and there are one wide peak at 130 and a smaller
  95   peak at 83, in addition to one peak at 0.
  96
  97 <h3>Image enhancement</h3>
  98
  99 We can get a good contrast stretch by using the histogram
 100 equalisation.  This will give us the widest range of visible
 101 separation between features.
 102
 103 <h3>Displaying colour images</h3>
 104
 105 <p><img width="40%" src="http://home.online.no/~oe-aase/jotunheimen/jotun2000topper.jpg">
 106 <!-- img src="jotunheimen-map.jpeg" -->
 107
 108 <img width="40%" src="jotunheimen-std-ir.jpeg">
 109
 110 <p>Comparing a map we found on the web, and the standard infrared
 111 image composition, we can identify some features from the colors
 112 used:</p>
 113
 114 <ul>
 115
 116  <li>water is black or green
 117
 118  <li>ice and glaciers are white, while snow is light green.
 119
 120  <li>vegetation is red.
 121
 122  <li>non-vegetation is brown or dull red when closer to snow and
 123    glaciers.
 124
 125 </ul>
 126
 127 <img align="right" width="40%" src="jotunheimen-ir-2band.jpeg">
 128 <p>Next, we tried to shift the frequencies displayed to use blue for the
 129 red band, green for the near ir band and red for the mid ir (1.55-1.75
 130 um).  With this composition, we get some changes in the colours of
 131 different features:
 132
 133
 134 <ul>
 135  <li>water is black
 136
 137  <li>ice and glaciers are light blue, while snow is dark blue.
 138
 139  <li>vegetation is light green and yellow.
 140
 141  <li>non-vegetation is red or brown.
 142
 143 </ul>
 144
 145 <h2>Filtering and image sharpening</h2>
 146
 147 <p>We decided to work on the grey scale version of the thermal infrared.
 148 This one has lower resolution then the rest of the bands, with 120m
 149 spatial resolution while the others have 30m spatial resolution.
 150
 151 <p>The high pass filtering seem to enhance the borders between the
 152 pixels.  Edge detection gave us the positions of glaciers and water.
 153 We tried a gradient filter using this 3x3 matrix: [ 1 2 -1 / 2 0 -2 /
 154 1 -2 -1 ].  It gave a similar result to the edge detection.
 155
 156
 157 <p>We also tried unsharp filtering using this 3x3 matrix: [ -1 -1 -1 / -1
 158 8 -1 / -1 -1 -1 ].  This gave similar results to the edge detection
 159 too.
 160
 161 <p>We started to suspect that the reason the 3x3 filters gave almost the
 162 same result was that the fact that the spatial resolution of the
 163 thermal band is actually 4x4 pixels.  Because of this, we tried with a
 164 5x5 matrix, making sure it sums up to 0.
 165
 166 <p><table align="center">
 167   <tr><td>
 168     <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
 169     <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
 170     <tr><td>-1</td><td>-1</td><td>24</td><td>-1</td><td>-1</td></tr>
 171     <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
 172     <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
 173   </table></p>
 174
 175 <p><img align="right" width="40%"src="jotunheimen-therm-unsharp5x5.jpeg">
 176 Next, we tried some different weight:
 177
 178   <p><table align="center">
 179     <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
 180     <tr><td>-1</td><td>-2</td><td>-2</td><td>-2</td><td>-1</td></tr>
 181     <tr><td>-1</td><td>-2</td><td>32</td><td>-2</td><td>-1</td></tr>
 182     <tr><td>-1</td><td>-2</td><td>-2</td><td>-2</td><td>-1</td></tr>
 183     <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
 184   </table></p>
 185
 186 <p>This one gave more lines showing the borders between the thermal
 187 pixels.
 188
 189 From: shanette Dallyn <shanette_dallyn@yahoo.ca>
 190 Subject: Re: My notes from todays exercise
 191 To: Petter Reinholdtsen <pere@hungry.com>
 192 Date: Sat, 30 Apr 2005 15:16:59 -0400 (EDT)
 193
 194 Hey Petter!
 195  Allright, I looked up some stuff on statistics and the most valuable
 196 conclusion that I can come up with for the histograpm peak question is:
 197 "The peak values of the histograms represent the the spectral sensitivity
 198 values that occure the most often with in the image band being analysed"
 199
 200 For the grey level question go to http://www.cs.uu.nl/wais/html/na-dir/sci/
 201 Satellite-Imagery-FAQ/part3.html I found this and thought that the first major
 202 paragraph pretty much answered the question for the grey levels.
 203
 204 theory of convolution:
 205
 206                       Specialty Definition: Convolution
 207
 208                    (From Wikipedia, the free Encyclopedia)
 209
 210 In mathematics and in particular, functional analysis, the convolution
 211 (German: Faltung) is a mathematical operator which takes two functions and and
 212 produces a third function that in a sense represents the amount of overlap
 213 between and a reversed and translated version of .
 214
 215 The convolution of and is written . It is defined as the integral of the
 216 product of the two functions after one is reversed and shifted.
 217
 218 The integration range depends on the domain on which the functions are
 219 defined. In case of a finite integration range, and are often considered as
 220 cyclically extended so that the term does not imply a range violation. Of
 221 course, extension with zeros is also possible.
 222
 223 If and are two independent random variables with probability densities and ,
 224 respectively, then the probability density of the sum is given by the
 225 convolution .
 226
 227 For discrete functions, one can use a discrete version of the convolution. It
 228 is then given by
 229
 230 When multiplying two polynomials, the coefficients of the product are given by
 231 the convolution of the original coefficient sequences, in this sense (using
 232 extension with zeros as mentioned above).
 233
 234 Generalizing the above cases, the convolution can be defined for any two
 235 square-integrable functions defined on a locally compact topological group. A
 236 different generalization is the convolution of distributions.
 237
 238 I hope this will help!
 239
 240 Shanette
 241
 242 <H2>References</h2>
 243
 244 <ul>
 245   <li><a href="http://www.cs.uu.nl/wais/html/na-dir/sci/Satellite-Imagery-FAQ/part3.html">Satellite-Imagery-FAQ</a>
 246 </ul>
 247
 248     <hr>
 249     <address><a href="mailto:pere@hungry.com">Petter Reinholdtsen</a></address>
 250 <!-- Created: Sun May  1 13:25:38 CEST 2005 -->
 251 <!-- hhmts start -->
 252 Last modified: Sun May  1 14:28:48 CEST 2005
 253 <!-- hhmts end -->
 254   </body>
 255 </html>