Quiz Four

Geo 352 Spring 2004

Jonathan Aguero

Section One: Contrast Enhancement

Raw Image of San Diego, California .

Satellite Platform: Spot Sensor (Panchromatic mode).

Spatial Resolution: 10 m (Panchromatic mode).

Spectral Resolution: 1 band (0.51-0.73 mm).

Temporal resolution: 26 days nadir viewing,

1 and 4 (or occasionally 5) days off nadir viewing.

Radiometric Resolution: 6 bit Digital Pulse Code Modulation (radiometric accuracy of 8 bits or 256 DN)

Here the difference the actual limits and the current scale is very low: just 24 Digital Numbers in the lower tail and 1 DN on the upper tail.

Lower limit: 0 Upper limit: 255

Selecting “limits to actual” it results:

Stretching limits to actual basically 24 is moved to 0 so the image seems darker than the original one.

Lower limit: 24 Upper limit: 254

Then input limits to 99%

Removing 1% of the tails clearly enhance the image by providing more contrast to the medium values which hold more information about land features.

Lower limit: 33.92 Upper limit: 152.98

From here to the end of the section the limits remain constant.

Slicing the histogram in 46 virtually disappear the land features (white) and highlight the ocean features.

Now inverting the slice we obtained a virtually negative of the last image where the land features appear gray and the ocean appears white.

With a Histogram Equalize enhance both peaks of the histogram were stretched in such a way that the contrast was enhanced in the regions with more information.

Here a 3 slices density/level slicing contrast enhancement was performed, clearly there are present just three colors white, gray, and black.

Again a density/level slicing contrast enhancement but with 10 slices, this provided an almost continuous scale at least for the human eye that can not distinguish much more than 10 levels of gray, especially the darkest ones levels.

Section Two: Using Formulas

Sub-section One

The first formula cannot be applied because there is just one band in the image so there cannot be an input3 object at least you add some other band/image to the display.

For the second formula we have:

Lower limit: -76 Upper limit: 154

The image is almost black because you are subtracting 100 to all the values of the image but keeping the limits between 0 and 255 so very few values depart from black (or very dark gray).

Sub-section Two

Before stretching limits to actual and performing a 99% contrast enhancement the image is completely black because all the values are between cero and 4.85. The image is not presented due to that.

The resulting image is the ratio between band four and band three, which is some kind of vegetation index. The most important vegetation areas are very bright in the image.

The resulting image is the Normalized Difference Vegetation Index for Landsat images. It is used for monitoring vegetation.

Both indexes differentiae vegetation from the rest but the NDVI makes a better job by increasing the contrast in the densest areas of the histogram.

First Principal Component: It shows most on the variability in the image. Two dominant peaks are present, the black (ocean) and the middle gray (suburban maybe).

Second Principal Component: This presents just one peak and it is very prominent. The Formula it is not present for sake of memory and disk space.

Third Principal Component: The peak is even more prominent than the last PC. The 99% of the values range from 13 to 79.

Fourth Principal Component: The peak is the most prominent one. The 99% of the values range from 24 to 43.

Section Three: Spatial Filtering

Spatial filtering is considered a local operation because it involves a spatial region instead of a single cell.

The resultant image is smoother than the original one because we are using a low-pass filter (3x3). Also the peaks are higher than in the original histogram.

The image is smoother than the former because we are using a bigger low-pass filter (5x5). Also the peaks are higher than in the last histogram.

This low-pass filter (7x7) is bigger than the last two so it produces an even smoother image and higher peaks in the histogram.

Section Four: Band Combinations

red = 3, green = 2, blue = 1

This band combination is known as true color because it recreates approximately the real color of the scene. The vegetation looks green, the pavement looks gray, and uncover soil appears brown.

red = 4, green = 3, blue = 2

Here the false color image highlights the presence of vegetation by using the band 4 (near infrared) in the red channel.

red = 5, green = 4, blue = 3

Here the false color image highlights the presence of vegetation by using the band 4 (near infrared) in the green channel. In addition the red band is presented in the blue channel which results in a yellow color for the most dense vegetation areas.