Stephen Teet Lab 5 Part 1 1. Landsat 4-5 TM (from Landsat Archive). 2. Rio Negro Basin / Manaus, Brazil (centered around -2.95,-60.58). September 2009. 3. Cloud cover = 0%. 4. Part 2 5. ENVI is using a linear method to stretch the image. 6. The breakpoints are at: Red = 14 & 52 Green = 19 & 44 Blue =57 & 86. 7. The image bacame darker when I removed the break points. The spectral bands of the image(red, green and blue) were being stretched to fill the available 8 bit range. This gave more brightness to the image because each band was given more display range. When the breaks were removed, the spectral bands were no longer being stretched (allowing room for the non visible bands which are not being displayed). 8. The equalization stretch gave the image more contrast (darker greens, more shades of blueviolet in the river). 9. Looking at the histogram, the equalization method stretches the image more in the parts of the spectrum that represents more pixels (the peaks) so that more detail can be seen where more data is available. Comparing this image to the first one, it appears that this stretching causes more represented spectral bands (green mostly, blue some in this case) to be displayed in the less colors of the less represented band (red). 10. Thegoal of the gaussian stretch seems to be to stretch available spectral data to fit a gaussian curve that stretche the full range of the 8 bit display. The gaussian stretch softens the image (less bold), but more similar to the equalization stretch in color representation. 11. Thehigh pass filter enhances areas with big contrast, such as the boundary between a forest and baresoil/concrete. The low pass filter enhances most regions of contrast. The high pass filter diminishes contrast in more homogeneous regions (i.e. within forested region). The low pass filter does not appear to diminish features much. 12. The high pass filter would be useful in finding boundaries between different land usage areas. The low pass filter would be more usefule in looking at differences within classes such as differing vegetation types. 13. Changing the kernal size to 5x5 made the image less sharp, this is because it smoothed the image using more surrounding pixels. 14. 15. 16.