Image classification

Question 1

When is unsupervised classification used?

Answer 1

When there is insufficient observational evidence of land cover

Question 2

What are the steps in unsupervised classification?

Answer 2

1. Select band
2. Select a cluster algorithm
3. Determine information classes corresponding to each cluster
4. Combine /eliminate
5. Assess results and make changes

Question 3

Compare supervised and unsupervised image classification

Answer 3

Supervised: training sites may be inadequate, time consuming/tedious, only pre defined classes are found
Unsupervised: clusters may be unidentifiable, time consuming and tedious, unexpected categories may be revealed

Question 4

What are the types of unsupervised classification algorithms?

Answer 4

K means(re evaluates means), isoclust(3 band color composite image to begin cluster seeding) isodata, cluster(peak histogram method)

Question 5

Compare hard and fuzzy assignments

Answer 5

Hard assigns each pixel one class whereas fuzzy gives each pixel a grade value for each class then has additional logic to membership grade information to derive final classifications

Question 6

What is supervised classification?

Answer 6

The user determines information classes to be identified

Question 7

Steps for supervised image classification

Answer 7

1. Decide on information classes
2. Choose training areas for each class
3. Statistical analysis
4. Approaches to band or channel selection
5. Select classification algorithm
6. Classify entire scene
7. Assess results and make necessary changes

Question 8

What is unsupervised classification?

Answer 8

When image is broken into spectral classes based on natural groupings in the data