projection method in order to transform two dimensional data into one dimensional data

Dear all, I m having a 2D feature space between two similar and dissimilar image. i want to construct a decision boundary to classify this feature or also want to apply projection method that can tranform the data into one dimensional feature space. can anybody suggest me any solution or method in matlab regarding this. I have done alot of research regarding this but, yet not get any solution. please help me if anyone have any idea of this. thank alot .

Answer 1

[Walter Roberson]

One way would be to calculate the "support vector", and having calculated it, rotate the data so that the support vector was perpendicular to the x axis; having done that, anything that has an x lower than the place the support vector projects to would be one class, and anything higher would be the other class.

Answer 2

[Promila]

thanks alot for reply.

is there any way to calculate the support-vector in matlab?

[Walter Roberson]

See http://www.mathworks.com/help/toolbox/bioinfo/ref/svmtrain.html if you have that toolbox, and if not search Answers for "libsvm"