From: LFG Tech on
Hello Friends,

I am trying to extract pixel intensities along the normal directions to a point.

I have an amoeba shaped body , in a binary image. The body is white(1) while the background is black. I have traced all the co-ordinates along the boundary of this body. Now for each point on the boundary , i want to extract pixel intensities along the normal directions . This would mean 2 directions - 1) one away from the body and towards the background and 2) into the body .

I have a vague idea that gradients give information about the normals/perpendiculars. Could you pls direct me to any site pls?

Regards,
LFG.
From: Ashish Uthama on
#1 How would you define a perpendicular line to a point conceptually?

Maybe you could pick N points on either side of the current point and fit
a straight line to these 2N+1 points. You could then obtain the equation
of the line perpendicular to this fitted line segment passing thought the
current point.

Checking the value of the first T points along either direction should
help you decide if you are going away or towards the body. Since you are
working on a mask (logical/binary image), you'll be getting either 0's or
1's. You could then use these indices to obtain pixel intensities from the
original image.

This might work for most points except really convoluted parts (Where both
directions of the normal might pass thorough the body at some point).



On Wed, 16 Jun 2010 00:58:03 -0400, LFG Tech <vinodkaruvat(a)gmail.com>
wrote:

> Hello Friends,
>
> I am trying to extract pixel intensities along the normal directions to
> a point.
>
> I have an amoeba shaped body , in a binary image. The body is white(1)
> while the background is black. I have traced all the co-ordinates along
> the boundary of this body. Now for each point on the boundary , i want
> to extract pixel intensities along the normal directions . This would
> mean 2 directions - 1) one away from the body and towards the
> background and 2) into the body .
>
> I have a vague idea that gradients give information about the
> normals/perpendiculars. Could you pls direct me to any site pls?
>
> Regards,
> LFG.