Re: About BIJ PCA usage

Posted by Joris FA Meys on Mar 15, 2009; 7:50pm
URL: http://imagej.273.s1.nabble.com/About-BIJ-PCA-usage-tp3693307p3693310.html

Dear,

without having looked at the code : if it is a plugin, chances are that the
access is not very straight-forward as most plugins are not written with
that access in mind. In case you don't find your answer here, just study the
code of the plugin and eventually adapt it for allowing better access. You
won't be the first one to do that.

Kind regards
Joris

On Sun, Mar 15, 2009 at 12:07 PM, Xin ZHOU <[hidden email]> wrote:

> Gabriel Landini a écrit :
>
>  On Saturday 14 March 2009, Xin ZHOU wrote:
>>
>>
>>> I would like to use the PCA transformation to reduce the dimension of
>>> the features.
>>> However, even it exist PCA components in BIJ plugins, no documents are
>>> gieven.
>>> Can some one give an example of how to use it?
>>>
>>>
>>
>> You submit a stack of images. Each slice represents a different variable.
>> So for pixel (x,y) across the stack, the z value are the variables
>> associated to it.
>> The result should be another stack in which each slice is a component. You
>> also get a Eigenvalue spectrum plot which shows the amount of the total
>> variance explained by each component.
>>
>> I hope it helps
>>
>> G.
>>
>>
>>
> Hello,
>
> First of all, I didn't represent a image as a vector, but a feature.
> Secondly, I'm not looking for the theory, but an implementation, like :
>
> a is the set of vector and c is the result,
> the function is c=new PCA(a) ......
>
> Because in the javadoc I didn't find the entry point for this class at all,
> so I'm looking for how to use it.
>
> Yes, by the way, I'm programming with ImageJ, so not just look for a button
> to click on :-)
>
> cheers, Xin
>