Lecture 9 - Unsupervised classification

Question 1

Why do we produce land cover maps?

Answer 1

Land cover interacts strongly with water cycle, carbon cycle and climate system

Question 2

What is image classification?

Answer 2

Groupings of pixels having similar values

Question 3

Feature space plot

Answer 3

Scatter plot between different wavebands most often 2D

Question 4

Euclidean Distance

Answer 4

Method to find distance between points and cluster centre

Question 5

Advantages of Digital classification over visual interpretation

Answer 5

• Cost-efficient in analyses of large data sets
• Results can be reproduced
• More objective than visual interpretation
• Effective analysis of complex multi-band spectral interrelationships

Question 6

Unsupervised classification

Answer 6

Classification method which classifies pixels with no prior information

Question 7

What is an unsupervised classification based on?

Answer 7

• Based on spectral groupings
• Requires interpretation after classification

Question 8

Two Methods?

Answer 8

ISODATA - calculates class means evenly distributed in the data space then iteratively clusters the remaining pixels using minimum distance techniques.
K - Means - calculates initial class means evenly distributed in the data space then iteratively clusters the pixels into the nearest class using a minimum distance technique.

Question 9

Advantages of Unsupervised

Answer 9

• No extensive prior knowledge of area needed
• Opportunities of human error are minimised
• Unique classes can be identified