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