Enhancing and Viewing Landscape Images

1
Enhancing and Viewing Landscape Images
2
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.

3
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.

4
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.

5
Contrast Manipulation

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

6
(No Transcript)
7
Contrast Manipulation
8
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.

9
(No Transcript)
10
Contrast Manipulation

• Normalize, or Gaussian stretch
• Forces the data into a bell-shaped curve.
• Tendency to spread the tails of the histogram.
  Used when dark areas or light areas need to be
  seen better.

11
(No Transcript)
12
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.

13
(No Transcript)
14
Contrast Manipulation

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

15
(No Transcript)
16
Contrast Manipulation

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

17
(No Transcript)
18
Contrast Manipulation
19
Contrast Manipulation

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

20
(No Transcript)
21
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 - - -

22
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.

23
Spatial feature manipulation
24
Original Low High
25
5,4,3 5,4,3(High Pass)
26
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.

27
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 and band 3 of Landsat.

28
(No Transcript)
29
Multi-image Manipulation-NDVI
30
Multi-image Manipulation-5/4