Prev: nonlinear curve fitting choices
Next: file browser inside the GUI
From: Deepa on 2 Jun 2010 16:13
I wish to apply two compression algorithms in a single image at a time. ie., The image has to be splitted into smooth and textured region based on standard deviation. (If stddev is 0, it stands for smooth and if not, textured). For smooth regions, one particular compression algorithm has to be applied and similarly for textured regions , another compression algorithm has to be applied. Can anybody help me in solving the problem plz....?
From: Walter Roberson on 2 Jun 2010 17:00
Deepa wrote:
> I wish to apply two compression algorithms in a single image at a time.
> ie., The image has to be splitted into smooth and textured region based
> on standard deviation. (If stddev is 0, it stands for smooth and if not,
> textured). For smooth regions, one particular compression algorithm has
> to be applied and similarly for textured regions , another compression
> algorithm has to be applied. Can anybody help me in solving the problem
> plz....?

stddev of 0 can only occur if the sum of the squares of the data minus the
mean is 0, which in turn can only occur if all of the data has a value that is
identical to the mean -- a constraint which can be reinterpreted as simply "a
region has stddev of 0 if and only if the region consists of identical values".

A region which is identical values compresses to a single copy of the value
together with information about the shape of the region.

Regions of identical values are not usually very extensive in real images, as
there is usually a fluctuation (even if only random) in values due to
differences in illumination, shadows, sensitivity of each point in the sensor
matrix, reflections, texture, and so on.

I'm looking out my window now and see nothing but clear blue sky without a
cloud at all, but if I were to take a picture of it, it would be full of
changes -- different angles towards the sky would represent differing amounts
of atmosphere for the light having traveled through, which is going to affect
both the brightness and the amount of blue scattering (which affects the color)...


Anyhow: sounds to me like you should be considering using logical indexing.
 | 
Pages: 1
Prev: nonlinear curve fitting choices
Next: file browser inside the GUI