Quiz 1

Question 1

What is spatial data?

Answer 1

shape, size, orientation and texture.

Question 2

What is spectral data?

Answer 2

Analyst’s knowledge and experience with the remote sensing data

Question 3

What is temporal data?

Answer 3

Crop phenology, vegetation phenology

i.e., deciduous vs. evergreen vegetation

Question 4

What is atmospheric absorption?

Answer 4

Particles and gases in the atmosphere can affect EM radiation as it comes into and leaves the atmosphere. These effects are caused by the mechanisms of scattering, absorption, and refraction.

The spectral ranges of remote sensing are always carefully chosen to avoid
atmospheric absorption. Therefore, it is less of a problem than scattering
to remotely sensed imagery.

Question 5

What are examples of atmospheric correction methods?

Answer 5

1. Dark Object Subtraction (DOS)
2. The Empirical Line Method
3. Image Regression Approach
4. More advanced absolute atmospheric correction

Question 6

What are some image classification approaches: computer-aided classification?

Answer 6

1. Unsupervised classification:
   the users let the computer to identify pre-specified number of spectral clusters among which the difference between clusters are maximized and within clusters are minimized.
2. Supervised classification:
   Users provide the computer with some examples (i.e., training pixels) of known features in multi-dimensional feature space.

Question 7

What are some examples of unsupervised classification?

Answer 7

Example: K-means, ISODATA (Iterative Self-Organizing Data Analysis Technique)

Question 8

What are some examples of supervised classification?

Answer 8

Example: Maximum Likelihood Classification

Question 9

What is maximum likelihood classification?

Answer 9

In a two dimensional spectral space,  each class form a probability density region,
 the probability for a pixel belongs to a center category is defined by its distance
 measured in standard deviation. The pixel is assigned a class that is most likely 
 (maximum likelihood).

Disadvantages: Large number of computations, particularly many spectral bands
are involved.

Question 10

What is an accuracy assessment?

Answer 10

If you close you eyes and randomly label the pixels in the image in one of the 7 classes, you will not be completely wrong. Kappa statistics accounts for this effect in your accuracy assessment.

Question 11

How do you improve classification accuracy?

Answer 11

-Appropriate classification scheme
-Better (and more) training data points
-Use multiple images (i.e., image stack)
-Try more advanced classification algorithms
(Neural network, CART, support vector machine)

Question 12

How does soft classification compare with hard classification?

Answer 12

Hard classifications - each pixel belongs to the class it most closely resembles

Soft classifications - each pixel can belong to more than one class and has membership grades for each class

Question 13

What are some examples of change detection?

Answer 13

1) Image differencing
2) Image ratioing
3) Change Vector Analysis
4) Post-classification comparisons

Question 14

What is image differencing?

Answer 14

Image differencing
The most straight forward way to see whether a change has happened is to take a
difference between two images collected over the same place at different times.

Question 15

What is image ratioing?

Answer 15

Image ratioing for change detection is based on the following fact that the ratio
of the DN values for a stable feature over two dates would be unity, while changed
pixels would have a ratio significantly different from unity.

However, due to various external factors, the stable features may not have an
expected value of unity from the ratio. A critical step is to determine the threshold
value to differentiate change/no-change, which is often empirical.

No from-to information on change detection is available.

Question 16

What is change vector analysis?

Answer 16

The reflectance values collected for a pixel can be thought of as coordinates for
a vector in multidimensional space. The vector difference between two dates
indicated the change for a pixel.

Change can be measured in two ways, the magnitude of change (the modulus of the
change vector) and the direction of change (the angle between the two vectors).

Question 17

What is Post-classification comparison?

Answer 17

Advantage: easy to implement and interpret “from-to” information
Disadvantage: error accumulation

Question 18

What is Dark Object Subtraction?

Answer 18

Assumption – Each scene should
have some pixels with “close to zero” brightness value.

DN = 0 if no atmospheric effects

Dark objects here - clear water

DOS is perhaps the simplest yet most widely used atmospheric correction method

Question 19

What is the Empirical Line Method?

Answer 19

. “two spectrally uniform targets in the image, one dark and one
bright, are identified”

2. “their actual reflectances are then determined by field”
3. Apply linear regression techniques

Question 20

What is image regression?

Answer 20

Use all the pixels in the images to do a band to band regression

Advantages: simple, straight forward
Disadvantages: Assumes no change in the image. Ironically,
the atmospheric correction is often done for change detection!

Lots of image processing work, contains arbitrary decision.

What about phenology?
If you cannot solve, avoid! Use image on anniversary date might help