EX2 - ImageRation/Change Detection Flashcards
What is Image Ratio / Band Ratio
quotients between measurements of reflectance in separate portions / bands of the spectrum
map land covers reflect high in one band and low in another band
two features have different spectral responses, the ratio of the two provides a single value that expresses the contrast between the two to differentiate between the features
If two features have the same spectral behavior, ratios provide little information
Image ratios utilize spectral information, but display the information as ratio maps
Which bands are useful for image ratioing
NIR: both soil and vegetation are medium bright. Hence, the NIR band is not very useful for differentiating between vegetation and soil
NIR / RED ratio
the ratio value for vegetation is much larger than the ratio value for soil
using ratio values, one can distinguish between the two
The two main purposes of image ratioing
Differences between the spectral reflectance curves of surface types can be brought out, and surface types can be identified
Aids image interpretation by removing the effect of topographic and shadow effect. Illuminated and non-
illuminated areas of the same surface type will have different BVs, but when ratioed, the ratios values will be same
How does image ratio remove topographic / shadow effect
Shadows have a scaling effect on BVs:
The reflective properties of surface features remain the same between wavelengths but the total brightness is
diminished
Reflection between different wavelengths are not altered but they the scale of total brightness
How does image ratio help identify certain surface types
enhance or reveal latent information of inverse relationship between two spectral responses to the same phenomenon
Common example of image ratio (NIR/R):
Slides 4 and 11
From the general spectral reflectance the following observation can be made:
Vegetation: NIR/R»_space;> 1,
Water: NIR/R < 1
Soil: NIR/R > 1
NIR/R
images can serve as a crude classifier of images, and indicate vegetated areas in particular. Therefore this ratio has been developed into a range of different
vegetation indices.
Creating and Interpreting band ratios
Band Ratio =
BV row 1, column 1, band X/BV row1, column 1, band Y
What are the range of possibilities for 8 bit image
255 / 1high = 255
1/ 255 low = .003921
Depending on the brightness value, the equation being used and depth of the image the ratio, the ratio may be greater of lower proportionally.
The Normalized Ratio NDVI: Full form; bands used; formula; value ranges; application(s)
(Normalized Differenced Vegetation Index).
developed specifically for vegetation,
can be used for anything.
VNIR - RED/ VNIR + RED
LANDSAT TM: (TM4-TM3)/(TM4+TM3)
Band Ratio =
Band X BV – Band Y BV
/
Band X BV + Band Y BV
Data would range from 1 to 1, in real numbers. Any place where reflection was equal in both bands would be coded 0
1 (Band X bright compared to Band Y)
-1 (Band X dark compared to Band Y)
0 (Equal reflection in both bands)
Generally, NDVI histogram shows 2 peaks One for pixels bright in Band X relative to Band Y; the other for pixels that are bright in Band Y relative to Band X
What is change detection?
analysis technique that uses repeated imaging to assess changes in surface features
Bi - temporal change detection:
Multi-temporal change detection / time series analysis:
changes detected by comparing the spectral changes in reflectance that may have occurred between different time periods
Good practices for change detection analysis
9 reasons
must be comparable
Acquired from same or similar sensor
Acquired at same time of day
Same IFOV and look angle
Acquired in same season
co-registered (geometric correction)
cloud free in the area of analysis
corrected for atmospheric effects
Change Detection techniques
c. Post Classification Comparison
each imagery is independently classified to fit common land types / land covers.
Land cover maps from the two time periods are overlaid and compared pixel by corresponding pixel
The result is a map of land type / land cover change
This is very useful when we not only want to detect
change but also want to attribute change
Pros: It can attribute changes; Easy to quantify area and rate of change
Cons: Influenced by classification accuracy
Application areas of change detection analysis
Land cover changes
Deforestation
Desertification
Urban growth / urban sprawl
Disaster monitoring
Agriculture
Vegetation health
Coastal change
Environmental impact assessment
Change Detection techniques
b. Image Algebra
Image differencing;
Image ratioing;
Euclidean Distance;
Change Vector Analysis)
Change Detection techniques
a. Visual Interpretation
First place to start, Visually compare images by:
Display one image on top of the other and then visualize change by swiping
one image over the other or flickering between images
Display the two images side by side, using the typical pan, zoom and cursor
inquiry tools for targeted analysis
Create a multi-date color image composite and (that is, all bands from both dates are stacked to form one image). Then display bands from different dates to highlight changes
Challenges:
Time consuming
Subjective
The Normalized Ratio NDVI: application(s)
•RED / BLUE
•TM 3 / TM 1
This ratio distinguishes between limonitic and non limonitic rocks
• GREEN / RED
• TM 2 / TM 3
This ratio also discriminates ferric iron content
High green to red ratio = vegetation and non limonitic rock or soil = right side of histogram
Low green to red ratio = limonite reddish rock and soil
= left side of histogram
In a ratio image the black and white extremes of the
gray scale represent pixels with the greatest
difference in reflectivity between the two Spectral
bands.
The darkest signatures are areas where the denominator of the ratio is greater than the
numerator.
Conversely the numerator is greater than the
denominator for the brightest signatures.
Where denominator and numerator are the same,
there is no difference between the two bands
Red materials, such as the Chugwater
Formation with its high content of iron oxide, have their maximum reflectance in band 3. Thus, in the ratio image 3/1 (red/blue) of Figure A, the Chugwater outcrops have
very bright signatures
