Image Classification

Question 1

Goal of classification process?

Answer 1

A key goal of any classification process is simplification or a reduction in the complexity of a system into something that is meaningful to the observer

Question 2

Classification Algorithms:
 Minimum Distance to Mean

Answer 2

Relies on the straight line distance from the class means to the unclassified pixel

• All of the pixels in the image are assigned to one of the specified classes
• Insensitive to different degrees of variation in the spectral response of data

Question 3

Classification Algorithms:
 Parallelepiped

Answer 3

• Specifies a minimum and maximum range of values for each class
• Range results in a rectangular box known as a decision region
• The advantage is that a small number of classes can be specified and some of the image left unclassified

Question 4

Classification Algorithms:
 Stepped Parallelepiped

Answer 4

Boundaries developed based on a stepped approach which allows us to create tighter “boxes” to define the classes.

Question 5

Classification Algorithms:
 Maximum Likelihood Classifier

Answer 5

• Evaluates both variance and covariance of the category spectral response pattern
• Calculates the probability (statistical) of a pixel belonging to each class

Question 6

Non - Parametric Classification: Spectral Mixture Analysis (SMA)

Answer 6

• one pixel can reflect more than one element or class

- a Spectral Mixture analysis decomposes a pixel into its proportions

Question 7

Non - Parametric Classification: SAM - Spectral Angle Mapper

Answer 7

The SAM is a nonparametric classifier that uses the shape of a spectrum (or any other linear combination of attributes) as a distinguishing criterion.

Question 8

Non - parametric Classifier: Support Vector Machine (SVM)

Answer 8

a form of supervised classification technique that classifies an unknown sample into a predetermined class.
- the data points are mapped with a projection that will maximize their separation