Prev: inspect enumerations
Next: Save structure to .mat file from C
From: Prashant sangulgi on 11 Aug 2010 15:13
Means to find variance of the matrix i have covert my image into gray level image then we'll get the variance of the image...... Is it ?
From: Walter Roberson on 11 Aug 2010 15:18
Prashant sangulgi wrote:
> Means to find variance of the matrix i have covert my image into gray
> level image then we'll get the variance of the image...... Is it ?

No, that would give you the variance of the grayscale image, not the variance
of the original image.

Is your image an indexed image (2D array of integer values) or is it a
true-color image (3D array of either integer values or double precision values
in the range 0 to 1) ?

What do you intend to do with the variance of the image once you know it?
Perhaps there is an alternative approach that does not require the actual
variance, or perhaps it makes sense to do the variance on a channel by channel
basis.
From: Prashant sangulgi on 11 Aug 2010 15:29
I am working on true color image.... How to find the variance and covariance when the image is true color image and is not graysale image....
From: ImageAnalyst on 11 Aug 2010 15:41
On Aug 11, 3:29 pm, "Prashant sangulgi" <psangu...(a)gmail.com> wrote:
> I am working on true color image.... How to find the variance and covariance when the image is true color image and is not graysale image....
---------------------------------------------------------------------------------------------------------------------------------------------
You tell me. No, let's take a step back and ask what it is that you
REALLY want to do. Don't give us some algorithm that *might* do what
you think you want to do, such as calculating the variance of your
image or subtracting something from your image or some other strange
thing presented without context. I'm asking you to post a picture
somewhere then tell us what you want to find or measure in the image.
Then WE will suggest the best MATLAB method for doing that. It may or
may not involve calculating the variance of one or more color planes.
From: Walter Roberson on 11 Aug 2010 15:47
Prashant sangulgi wrote:
> I am working on true color image.... How to find the variance and
> covariance when the image is true color image and is not graysale image....

variance has a natural extension multi-dimensional matrices in that one can
ask about the variance of the values all considered together,

var(A(:))

However, covariance implicitly asks the question of how one value correlates
to another value, so unless one is asking about a single correlation value
between all of the image values,

cov(A(:))

then one must define the axes that the variation is to be considered with
respect to. With a 2D matrix, a natural mechanism is to correlate the X axes
versus the Y axes, but with a 3D matrix, one needs to consider whether one
wants to correlate (X,Y), (X,Z), or (Y,Z)

What is it that you are attempting to discover or measure? Describe that in
your own words, because if you just repeat "variance" and "covariance" again
then I will assume that you have been given this as an assignment and that you
don't understand what the assignment is about.
 | 
Pages: 1
Prev: inspect enumerations
Next: Save structure to .mat file from C