RE: Land Use /Land Cover, Image classification

Dear Mr. Uday,

As I understand the focused problem is:

“How to take only vegetation classes such as agriculture land, forest, grass land etc., from supervise classification method & non-vegetation classes from unsupervised classification method to make a single final result.”

Answer: As per my knowledge it is possible. Please go through the attached .pdf file, where two classes (one from supervised and another from unsupervised) are exemplified. In the same way you may put more than two classes to create a single final result.

Regards

Partha Pratim Ghosh

PCI Software Pvt. Ltd.

Kolkata. India

+91 94325 84408

From: Uday J [mailto:uday.j@xxxxxxxx]
Sent: Friday, January 23, 2009 6:40 AM
To: PCI-GEO
Subject: Land Use /Land Cover, Image classification

Hi Dear All,

Sub: Regarding Land Use /Land Cover, Image classification.

As per our requirement we need to do Image classification for some of the visual classes as 01.Water,02.Forest,03.Urban Area etc., but according to the our exercise as per still now we have done classification of image from both method as supervise classification & unsupervised classification with using PCI Geomatica 10.1.3. But till now we have not got appropriate result what we are expecting. I want to notify you in ERDAS IMAGINE we have one more method to enhance classification result called as"Knowladge-Base-System",in this method we can apply supervised classification (maximum like hood method) to classify conditions in land use/land cover and then join spectral characters and GIS data by knowledge-based system for increasing classification accuracy and categories.

In simple words, its very difficult to collect non-vegetation signature set like as urban, fallow land, degraded land etc., while doing supervise classification, but we can get nice result of non-vegetation classes at unsupervised classification method, so if we required to take only vegetation classes such as agriculture land, forest, grass land etc., from supervise classification method & non-vegetation classes from unsupervised classification method to make a single final result, then above mentioned("Knowledge-Base-System") is very much useful for same. This method is works on "Knowledge Engineer" tool base in ERDAS IMAGINE to get combination for classes, which has obtained from two different method (Supervise/Unsupervised) according to the priority given user called as confidence level of the class.

Herewith, I am requesting you to please let me know if there is such kind of method is available in PCI Geomatica to enhance image classification result, then please try to send me the process flow and methodology for same.

Please, feel free if you have any question is there about this, and in case if you have any idea about the similar methodology from PCI Geomatica then please send me processing steps and related tutorials for more reference for me.

Waiting for yours kind reply as soon as possible.

Thanks & Regards

Uday

GIS Remote Sensing Specialist

SECON TECH PARK

SECON SURVEY (P) LTD.,

“A GIS driven Multidiscipline Company”

Plot No.147, 7 B Road, EPIP Zone.

Whitefield, Bangalore-560 066.

India.

I Total Solution From a Single Window I

Go to: www.secon.in