ES 555 Lab Exercise
Digital Resampling

Lake Kahola, Kansas
James S. Aber


Introduction

This exercise is based on two kite aerial photographs from Lake Kahola in the Flint Hills of east-central Kansas. The images were acquired in February, 2002, during the winter, leaf-off season. The vertical photographs are similar in scale and orientation, and have substantial overlap in ground coverage. The two photographs are digital images, measuring 1200 by 1600 pixels, and named kahola1 and kahola2. Transfer these jpg files via FTP into your personal computer work space.

For further examples and background information
about the geographic setting--see Lake Kahola.

The exercise involves three software packages for different phases of the image processing.

  1. D Joiner -- Designed for stitching together photographs in mosaic and panoramic views. From D Vision Works.

  2. Adobe Photoshop -- Industry standard software for manipulating and enhancing photographic images. See Photoshop.

  3. Idrisi Andes -- Most widely licensed raster-based GIS software package globally. From Clark Labs.

Note: This exercise is based on D Joiner, available in the Geospatial Analysis laboratory at ESU. Distance-learning students may utilize another image stitching program called PTGui. Trial software is available free for 30 days.

Exercise

Click on the D Joiner icon to start the software, and then maximize the working window to fill the entire monitor screen. Click on the file menu (upper left corner), then choose the "Insert Photos" option. Select kahola1 and kahola2, then click the "Open" button. You will next see a box titled "Matrix." Accept all default entries, and click OK. Now you should see the two images positioned side by side at the center of the window.

The two airphotos should appear in the arrangement portrayed above. Note: these are "raw" digital images as acquired by kite aerial photography using a Canon Digital Elph camera from a height of ~100 meters. The images have not received any type of enhancement or modification. North is toward the left side of both photos.

You are now ready to enter control markers for stitching the two images together. Click on the "Marker mode" icon (middle of three colored icons). The images will appear enlarged, side by side. Move the cursor to each of the other icons to see what their functions are. Select the "Marker Tool" (+) icon. Notice the low stone wall (white line) next to the red truck on the left sides of both images. Use the marker tool to place "stitches" at both ends of the stone wall. After placing these stitches, click the "Preview mode" icon. You should see a rough display of the stitched image. Click the "Zoom in" icon a few times to enlarge the image, then click "Render" icon (farthest to right).

This initial stitching is relatively good, but there are a few small offsets--see boat dock with yellow diving board. Having achieved this stage of operation, you should save the result (third icon from left). Enter an appropriate file name, which will be saved as a "dst" type file in the same folder as the input image files. There are a couple of ways to improve the fit of image stitching.

As you experiment with these techniques, note options for zooming, panning, moving markers, deleting markers, etc. Press the "F1" key for further information and suggestions. When you have achieved a good fit of stitched images, be sure to save your results.

Sample image of stitched airphotos. Your image should appear quite similar. Click on sample image to see larger version.

As a final step with D Joiner, export the image for further processing. Go to the file menu, and select the "Export, Flat" option. Enter an appropriate file name, such as kahola.jpg, and click the "Save" button.

Note: D Joiner geometrically distorts the images in order to join them together. The resulting concat image is not georeferenced.

Now continue the exercise using Adobe Photoshop (or equivalent) software to manipulate the image. Open the jpg image you exported from D Joiner. You should carry out several operations in this sequence.

When you have completed these steps, use the "Save as" option to convert your final result to a "bmp" file using an appropriate file name.

The last portion of this exercise involves Idrisi Andes software to apply geospatial data. Import the bmp image, as you did in previous exercises. Examine the metadata; note the data type and other attributes of the image.

1. How many rows and columns does the image have?

2. Why is resolution listed as unknown? Why are the value units listed as unspecified?

The most important task, from a GIS perspective, is to determine pixel resolution. Once again, examine the stone wall next to the red pickup truck. Enlarge the wall, so you can see individual pixels. Note the row and column positions for each end of the wall. Determining the length of the wall (in pixels) can be done using the Pythagorean theorem.

The wall is oriented diagonally to the row/column grid. To determine the size of individual cells is a simple right-triangle geometric problem, in which the wall represents the triangle hypotenuse. Rows and columns represent the other sides of the triangle. The relationship of the sides is given by the Pythagorean theorem, as modified below.

R² + C² = H²

Where:
R = difference in rows between ends of wall.
C = difference in columns between ends of wall.
H = hypotenuse or the length of the wall (in pixels).

Solve this relationship for length of the wall. Round your answer to the nearest tenth unit. The units of measurement are simply cells in the raster grid. The measured length of the wall is 17.2 meters. To determine the size of each cell, divide the wall length (17.2 m) by your result (H) from above. Round your answer to the nearest centimeter (0.01 m).

3. How many pixels long is the wall?

4. What is the pixel resolution?

Now update the metadata with appropriate information for resolution and for max x and max y values.

5. What values did you enter for max x and max y? How did you calculate these values?

6. How much ground area does the image cover? Give answer in m², hectares and acres.

7. What are the smallest objects you can identify? How would you rate the interpretability of this image? See NIIRS criteria.

Now think about orientation of the image. The wall is oriented 84° (toward the right end), or 84° east of true north. The north arrow for the image should, thus, point 84° counterclockwise from the east end wall. In order to place the north arrow, you need to know orientation of the wall relative to the raster grid of the image. Once again this is a simple problem in trigonmetry involving a right triangle. The angle of the hypotenuse (wall) is given as follows.

tan A = C ÷ R

Where:
A = angle of hypotenuse (wall) relative to columns.
R = difference in rows between ends of wall.
C = difference in columns between ends of wall.

Solve this relationship for angle of the wall. Round your answer to the nearest whole degree. The angle of the wall is relative to the image grid columns. This angle minus 84° is the image declination.

8. What is the declination for this image?

As your final task, construct a map composition that includes a title, scale bar and north arrow. Include your name and date as subtitles. Name your composition KAHOLA, and save a digital image file (bmp). Convert the bmp file to jpg or gif format to turn in.

Turn in

Return to SFAP schedule.
ES 555 © by J.S. Aber (2008).