Enhancing and Viewing Landscape Images

1
Enhancing and Viewing Landscape Images
2
(No Transcript)
3
Image Enhancement

• Procedures applied to image data to produce video
  displays or hard-copy output for visual
  interpretation.
• Goal Produce the single best image for a
  particular application.

4
Methods of Enhancement

• 1. Contrast manipulation
• Contrast stretching expand the DN values beyond
  their natural range to fill the 0-255 range.
• 2. Spatial feature manipulation
• Refers to image texture.
• Smooth areas have low spatial frequencies, gray
  values change gradually.
• Rough areas have high spatial frequencies and
  gray values change abruptly.

5
Methods of Enhancement

• 3. Multi-image manipulation.
• Two or more images combined mathematically,
  commonly by ratios.
• Used to develop green vegetative index images,
  e.g., the NDVI.

6
Contrast Manipulation

• Level slicing, density slicing, no stretch.
• Equal divisions of the 0-255 DN range with no
  change in data values.

7
(No Transcript)
8
Contrast Manipulation
9
Contrast Manipulation

• Linear stretch
• Input output with no change, or truncate low
  and high values to get better values across most
  of the histogram. Rarely occurring values get as
  much emphasis as common ones.

10
(No Transcript)
11
Contrast Manipulation

• Normalize, or Gaussian stretch
• Forces the data into a bell-shaped curve.
• Tends to saturate the tails of the histogram.
  Best used with normally distributed data.

12
(No Transcript)
13
(No Transcript)
14
Contrast Manipulation

• Histogram equalization
• Puts equal number of pixels at each gray level.
• Result is better contrast where most of the pixel
  values are in the histogram-the center.
• Lowest and highest DN values will be lost.

15
(No Transcript)
16
Contrast Manipulation

• Exponential
• One tail or the other is emphasized while
  everything else is depressed.

17
(No Transcript)
18
Contrast Manipulation

• Logarithmic stretch
• Similar to exponential in that one tail or the
  other may be emphasized.

19
Contrast Manipulation
20
Contrast Manipulation

• Multi-band image
• Each band is stretched independently.

21
(No Transcript)
22
Spatial feature manipulation

• Low pass filter
• Emphasize large area changes in brightness. Uses
  a 3x3, 5x5, etc., moving window over the data and
  calculates new pixel values.
• Example
• 67 67 72 - - -
• 70 68 71 - 70 -
• 72 71 72 - - -

23
Spatial feature manipulation

• Calculation is 1/9(67)1/9(67)1/9(72)
• 70
• Process called convolution.
• High pass filters emphasize areas of high spatial
  roughness.
• Accomplished by subtracting a low pass filter
  from the original image.

24
Spatial feature manipulation
25
Original Low High
26
5,4,3 5,4,3(High Pass)
27
Multi-image Manipulation

• Used to generate specific information desired by
  the analyst.
• What bands to use is predicted from spectral
  curves.
• Where do differences occur.

28
Multi-image Manipulation

• Look for bands that are not highly correlated.
• That show a wide spread of data in a scatterplot.
• For example, look at band 4 (nir) and band 3
  (red) of Landsat.

29
(No Transcript)
30
Multi-image Manipulation-NDVI
31
NDVI

• Normalized Difference Vegetative Index
• NDVI ?nir ?red / ?nir ?red
• Most scene values between 0 and 1, no live green
  plants to all live green plants.

32
Multi-image Manipulation-5/4