how to apply SVM algorithm for classification for multispectral remote sensing data?

i am going to separate various colors from the image using SVM so want to know how to start ?

Answer 1

[Image Analyst]

How many color classes do you have? Are you sure SVM is the best approach - why do you want to use it? I think it's mainly for two class situations but there is a multi-class extension to it.

I have some color segmentation demos in my File Exchange: http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862

[Priyanka Athawale]

dear sir , thank u for the reply !! m sure I want to go with SVM for multicolor(multiclass) classification on remote sensing image. but Am very confused from where to start? please if u have any idea plz help me for the same.the classes may be 3 , 4 , 5 because my aim is to classify various crops on Indian agricultural land .

[Image Analyst]

Sorry, I have not used multi-class SVM for color classification (yet). You will find color classification algorithms for remote sensing images here: http://iris.usc.edu/Vision-Notes/bibliography/contentscartog.html#Cartography,%20Aerial%20Images,%20Remote%20Sensing,%20Buildings,%20Roads,%20Terrain,%20ATR

Answer 2

[Yinguang Li]

You can go to the website http://www.mathworks.com/help/stats/support-vector-machines.html for more information, or type 'demo toolbox statistics' in your matlab command window suppose you have a statistical toolbox installed.

[Priyanka Athawale]

hello sir,

thank u for ur reply ..!! I have statistic toolbox , N got the option feature selection there . but how to proceed for SVM after that? and I have only one image , so how to check its accuracy , whether the classification is accurate or not??