I am trying to understand how Fuzzy classification works in ERDAS IMAGINE. The software provides an option for fuzzy classification. The user can select parametric rules such as maximum likelihood, minimum distance, and spectral correlation mapper and set fuzzy classification. Does the software really consider these parametric rules for classification?
Thanks in advance.
There are two parts to your question.
Firstly, the classification of satellite data (pixel wise) is carried out based upon the DN values attached which is due to the spectral signature of the content.
Secondly, the fuzzy system is constructed based upon the numerical values attached to pixel. Any above said algorithm can be used to build the numerical relation in an fuzzy system. Based upon these relations a set of rules are generated. This section where rules are generated can be set by the user or s/w may be allowed to choose automatically.
Fuzzy system makes use of the above values to generate an output which is closest to DN values.
@Sanjeev Gadad,
Dear Sir,
I am aware of the Knowledge Based and Decision Tree classifiers. If ERDAS creates a set of rules, how can we access the rules? Also, i am unable to find mathematical equations supporting fuzzy classification using above mentioned parametric rules.
Also, is it possible to measure the the values of membership functions assigned to the pixel under consideration by the fuzzy classifier?
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing