<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
- <title></title>
+ <title>Assigment 8 in GEG2210 2005</title>
</head>
<body>
Photogrammetry</p>
<h1>Image enhancement, filtering and sharpening</h1>
-<img width="40%" src="jotunheimen-std-ir-eq.jpeg">
<p>By Petter Reinholdtsen and Shanette Dallyn, 2005-05-01.</p>
<p>This exercise was performed by logging into jern.uio.no using ssh
and running ERDAS Imagine. Started by using 'imagine' on the command
-line. The images were loaded from /mn/geofag/gggruppe-data/geomatikk/
+line. The images were loaded from /mn/geofag/gggruppe-data/geomatikk/</p>
<p>We tried to use svalbard/tm87.img, but it only have 5 bands. We
decided to switch, and next tried jotunheimen/tm.img, which had 7
-bands.
+bands.</p>
+
+<h2>Some notes on the digital images</h2>
<p>The pixel values in a given band is only a using a given range of
values. This is because sensor data in a single image rarely extend
-over the entire range of possible values.
+over the entire range of possible values.</p>
<p>The peak values of the histograms represent the the spectral
sensitivity values that occure the most often with in the image band
-being analysed.
+being analysed.</p>
<h2>Evaluation of the different bands</h2>
-<p><img align="right" width="40%" src="jotunheimen-truecolor.jpeg">
+<p><img src="jotunheimen-truecolor.jpeg" align="right" width="40%">
This image show the "true colour" version, with the blue range
assigned to the blue colour, green range to green colour and red range
to red colour.</p>
peaks at 0 and 255. The mean value is 34.3403.
<h3>band 4, near-infraread (0.76-0.90 um)</h3>
-<img align="right" width="20%" src="jotunheimen-band4-hist.jpeg">
+<img src="jotunheimen-band4-hist.jpeg" align="right" width="20%">
Water acts as an absorbing body so in the near infrared spectrum,
water features will appear dark or black meaning that all near
<h3>Image enhancement</h3>
-We can get a good contrast stretch by using the histogram
-equalisation. This will give us the widest range of visible
-separation between features.
-
+<img src="jotunheimen-std-ir-lin.jpg" width="40%">
+<p>When we look at the linear contrast functions, we can move the
+slope and shift values increasing or decreasing the contrast of the
+image. For example, in the linear contrasting we moved the slope value
+from 1.00 to 3.00 to obtain a brighter appearing image, and then we
+moved the shift from 0 to 10 to recieve a sharper image.</p>
+
+<img src="jotunheimen-std-ir-piece.jpg" align="left" width="25%">
+<img src="jotunheimen-std-ir-pieceimg.jpg" align="right" width="40%">
+
+<br clear="all">Next we tried the piecewise linear stretching for
+contrast. In this image we tried to make all of the histograms in the
+red, blue and green spectrum as similar as possible so we could detect
+a change in the image.(insert histogram change)
+
+<img src="jotunheimen-std-ir-piece2.jpg" align="left" width="25%">
+<img src="jotunheimen-std-ir-pieceimg2.jpg" align="right" width="40%">
+
+<br clear="all">We tried to break the slope and move the break point
+to slightly after each histogram peak. This resulted in the image
+obtaining a slightly blue tint and dullness. (put ugly blueish picture
+here) As this result was not really increasing the contrast, we tried
+another variation to try to spread out the histogram peak to use a
+wider range. This setting gave an improved image, were it is easier
+to see the red vegetation and the white ice.</p>
+
+<br clear="all"><img src="jotunheimen-std-ir-eq.jpeg" align="right" width="40%">
+<p clear="all">We also tried to do histogram equilization on the
+standard infrared composition. This changed the colours in the image,
+making the previously green areas red, and the brown areas more light
+blue. In this new image, we can clearly see the difference between
+two kind of water, one black and one green. We suspect the green
+water might be deeper, but do not know for sure.</p>
+
+<p>We can get best contrast stretch by using the histogram
+equalisation. This gave us the widest range of visible separation
+between features.
+
+<br clear="all">
<h3>Displaying colour images</h3>
-<p><img width="40%" src="http://home.online.no/~oe-aase/jotunheimen/jotun2000topper.jpg">
+<p><img src="jotun2000topper.jpg" width="40%">
<!-- img src="jotunheimen-map.jpeg" -->
-<img width="40%" src="jotunheimen-std-ir.jpeg">
+<img src="jotunheimen-std-ir.jpeg" width="40%"></p>
<p>Comparing a map we found on the web, and the standard infrared
image composition, we can identify some features from the colors
used:</p>
+<img src="jotunheimen-ir-2band.jpeg" align="right" width="40%">
<ul>
<li>water is black or green
- <li>ice and glaciers are white, while snow is light green.
+ </li><li>ice and glaciers are white, while snow is light green.
- <li>vegetation is red.
+ </li><li>vegetation is red.
- <li>non-vegetation is brown or dull red when closer to snow and
+ </li><li>non-vegetation is brown or dull red when closer to snow and
glaciers.
-</ul>
+</li></ul>
-<img align="right" width="40%" src="jotunheimen-ir-2band.jpeg">
<p>Next, we tried to shift the frequencies displayed to use blue for the
red band, green for the near ir band and red for the mid ir (1.55-1.75
um). With this composition, we get some changes in the colours of
-different features:
+different features:</p>
<ul>
<li>water is black
- <li>ice and glaciers are light blue, while snow is dark blue.
+ </li><li>ice and glaciers are light blue, while snow is dark blue.
- <li>vegetation is light green and yellow.
+ </li><li>vegetation is light green and yellow.
- <li>non-vegetation is red or brown.
+ </li><li>non-vegetation is red or brown.
-</ul>
+</li></ul>
-<h2>Filtering and image sharpening</h2>
-<p>We decided to work on the grey scale version of the thermal infrared.
-This one has lower resolution then the rest of the bands, with 120m
-spatial resolution while the others have 30m spatial resolution.
+<h2>Filtering and image sharpening</h2>
-<p>The high pass filtering seem to enhance the borders between the
-pixels. Edge detection gave us the positions of glaciers and water.
-We tried a gradient filter using this 3x3 matrix: [ 1 2 -1 / 2 0 -2 /
-1 -2 -1 ]. It gave a similar result to the edge detection.
+<img src="jotunheimen-band4.jpeg" width="40%">'
+<p clear="all">We decided to work on the grey scale version of the
+near infrared (band4). We changed the colour assignment to use this
+band for all three colours, giving us a gray scale image.</p>
+
+<img src="jotunheimen-band4-low3.jpeg" width="40%">'
+<p clear="all">We applied the 3x3 low pass filter on this image, and
+this gave us almost the same image as the original. If you look
+closely you can see that some white dots in the original disapper, and
+some of the water edges seem to blur very slightly.</p>
+
+<img src="jotunheimen-band4-high3.jpeg" width="40%">'
+<p clear="all">We also tried the 3x3 high pass filter on the band4
+grey scale image. This gave a very noisy image. Edges of vallies and
+ice are not well defined. The black waters are still obvious.</p>
+
+<img src="jotunheimen-band4-edge3.jpeg" width="40%">'
+<p clear="all">We also tried the 3x3 edge detection, and this gave us
+an image that makes it difficult to distinguish elevation features
+such as the valleys. Rather, edge detection allows us to study main
+features in an area like the lakes. (insert band4 edge 3 image)
+
+<img src="jotunheimen-band4-grad3.jpeg" width="40%">'
+<p clear="all">We tried a gradient filter using this 3x3 matrix. The
+matrix was chosen to make sure the sum of all the weights were zero,
+and to make sure the sum of horizontal, vertical and diagonal numbers
+were zero too.</p>
+<p><table align="center">
+ <tbody><tr><td>1</td><td>2</td><td>-1</td></tr>
+ <tr><td>2</td><td>0</td><td>-2</td></tr>
+ <tr><td>1</td><td>-2</td><td>-1</td></tr>
+</tbody></table></p>
+
+<p>The gradient filter used gave us enhancement on lines in the
+vertical, horizontal and diagonal directions. This is seen by the
+white lines that outline certain areas of main features like the
+rivers within the vallies and some of the lakes.</p>
+
+<img src="jotunheimen-band4-neg1.jpeg" width="40%">'
+<p>When we rework the matrix to equal negative one, we end up with a
+lot of noise in the image that also seems to blurr the image. Using a
+negative one matrix is not optimal if you are trying to obtain
+sharpness.</p>
-<p>We also tried unsharp filtering using this 3x3 matrix: [ -1 -1 -1 / -1
-8 -1 / -1 -1 -1 ]. This gave similar results to the edge detection
-too.
+<p><table align="center">
+ <tbody><tr><td>-1</td><td>-1</td><td>-1</td></tr>
+ <tr><td>-1</td><td>7</td><td>-1</td></tr>
+ <tr><td>-1</td><td>-1</td><td>-1</td></tr>
+</tbody></table></p>
-<p>We started to suspect that the reason the 3x3 filters gave almost the
-same result was that the fact that the spatial resolution of the
-thermal band is actually 4x4 pixels. Because of this, we tried with a
-5x5 matrix, making sure it sums up to 0.
+<img src="jotunheimen-band4-plus1.jpeg" width="40%">'
+<p clear="all">We then tried with a 3x3 matrix were the sum of all
+values equals 1, to enhance the high frequency parts of the image.</p>
<p><table align="center">
- <tr><td>
- <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
- <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
- <tr><td>-1</td><td>-1</td><td>24</td><td>-1</td><td>-1</td></tr>
- <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
- <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
- </table></p>
-
-<p><img align="right" width="40%"src="jotunheimen-therm-unsharp5x5.jpeg">
-Next, we tried some different weight:
-
- <p><table align="center">
- <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
- <tr><td>-1</td><td>-2</td><td>-2</td><td>-2</td><td>-1</td></tr>
- <tr><td>-1</td><td>-2</td><td>32</td><td>-2</td><td>-1</td></tr>
- <tr><td>-1</td><td>-2</td><td>-2</td><td>-2</td><td>-1</td></tr>
- <tr><td>-1</td><td>-1</td><td>-1</td><td>-1</td><td>-1</td></tr>
- </table></p>
-
-<p>This one gave more lines showing the borders between the thermal
-pixels.
-
-From: shanette Dallyn <shanette_dallyn@yahoo.ca>
-Subject: Re: My notes from todays exercise
-To: Petter Reinholdtsen <pere@hungry.com>
-Date: Sat, 30 Apr 2005 15:16:59 -0400 (EDT)
-
-Hey Petter!
- Allright, I looked up some stuff on statistics and the most valuable
-conclusion that I can come up with for the histograpm peak question is:
-"The peak values of the histograms represent the the spectral sensitivity
-values that occure the most often with in the image band being analysed"
-
-For the grey level question go to http://www.cs.uu.nl/wais/html/na-dir/sci/
-Satellite-Imagery-FAQ/part3.html I found this and thought that the first major
-paragraph pretty much answered the question for the grey levels.
-
-theory of convolution:
-
- Specialty Definition: Convolution
-
- (From Wikipedia, the free Encyclopedia)
-
-In mathematics and in particular, functional analysis, the convolution
-(German: Faltung) is a mathematical operator which takes two functions and and
-produces a third function that in a sense represents the amount of overlap
-between and a reversed and translated version of .
-
-The convolution of and is written . It is defined as the integral of the
-product of the two functions after one is reversed and shifted.
-
-The integration range depends on the domain on which the functions are
-defined. In case of a finite integration range, and are often considered as
-cyclically extended so that the term does not imply a range violation. Of
-course, extension with zeros is also possible.
-
-If and are two independent random variables with probability densities and ,
-respectively, then the probability density of the sum is given by the
-convolution .
-
-For discrete functions, one can use a discrete version of the convolution. It
-is then given by
-
-When multiplying two polynomials, the coefficients of the product are given by
-the convolution of the original coefficient sequences, in this sense (using
-extension with zeros as mentioned above).
-
-Generalizing the above cases, the convolution can be defined for any two
-square-integrable functions defined on a locally compact topological group. A
-different generalization is the convolution of distributions.
-
-I hope this will help!
-
-Shanette
-
-<H2>References</h2>
+ <tbody><tr><td>-1</td><td>-1</td><td>-1</td></tr>
+ <tr><td>-1</td><td>9</td><td>-1</td></tr>
+ <tr><td>-1</td><td>-1</td><td>-1</td></tr>
+</tbody></table></p>
+
+<p clear="all">This gave us a sharper looking image compared to the
+result of the negative 1 filter. This is not really obvious unless
+one is comparing the two images carefully. In order to see more
+differences the matrix sums would have to be more then plus/minus one.</p>
+
+<h2>References</h2>
<ul>
<li><a href="http://www.cs.uu.nl/wais/html/na-dir/sci/Satellite-Imagery-FAQ/part3.html">Satellite-Imagery-FAQ</a>
-</ul>
+</li></ul>
<hr>
<address><a href="mailto:pere@hungry.com">Petter Reinholdtsen</a></address>
projects / homepage.git / blobdiff