Comments on: bwselect

By: Steve

Steve — Sun, 22 Mar 2009 22:30:25 +0000

Tarik—Use find to get the indices of the select object. Set the image elements at those location to zero. Offset the locations by the desired amount, and then set the image elements at the offset locations to be 1.

By: Tarik

Tarik — Sat, 21 Mar 2009 21:31:50 +0000

Please, if you can help me , how can I use bwselect function, to change the location of the object which we selected, for example,we want to give new coordinates for this object! is it possible?

By: Steve

Steve — Tue, 16 Sep 2008 11:58:02 +0000

Pete—Thanks for the kind words. I can’t think of a way to do what you suggest without getting into some custom GUI programming. See my GUI programming resources page.

By: Pete

Pete — Mon, 15 Sep 2008 18:00:38 +0000

Hi Steve. I have just recently started working with the image processing toolbox. At the moment I am trying to write a program that allows a user to select ‘fields’ on images of ordnance survey maps. I have been using bwselect and the mouse to allow the user to click on regions of interest. It would be very useful if I was able to display each selected region as colored after each single mouse click, as opposed to after the double click when all selections have been made. Is there any simple way to do this without creating a more complex GUI, or without creating a loop where the number of selections is known in advance? Alternatively, can you suggest some other suitable method? Thanks - great blog by the way!

By: Steve

Steve — Fri, 15 Feb 2008 13:28:46 +0000

Utk—You could use the interactive bwselect syntax, which lets you use the mouse to define starting locations, or you could use impixelinfo to display pixel location information on the figure window as you move the mouse around.

By: Utk

Utk — Fri, 15 Feb 2008 00:17:18 +0000

How do you go about guessing pixels fir the bwselect function. In the above example I downloaded the image and tried fiddling with the numbers and i couldnt see anything.

By: Steve

Steve — Tue, 29 Jan 2008 01:34:50 +0000

StevenB—The function normxcorr2 doesn’t support that behavior. You’ll have to modify the function yourself.

By: StevenB

StevenB — Mon, 28 Jan 2008 20:28:52 +0000

I’ve been playing with normxcorr2, specifically the image registration example at http://www.mathworks.com/products/demos/image/cross_correlation/imreg.html.

It says “The image sub_onion will be the template, and must be smaller than the image sub_peppers.”

But when you run the example, the matrix of outputs is huge. How can I make the function only compute the coefficients for the cases where the template fits within the image to which it is being registered.

For example, if I have a 100 X 100 master image and a 90 X 80 subimage, I want only a (100-90)X(100-80) = 10 X 20 matrix of coefficients. How do I tell the function not to pad the edges?

By: Steve

Steve — Mon, 07 Jan 2008 14:46:03 +0000

Clang—You might want to look at the color segmention demo. Click on the “Demos” link, under Image Processing Toolbox on the right side of the page.

By: Clang

Clang — Mon, 07 Jan 2008 02:33:04 +0000

Hi, im still a newbie at this. I’m trying to study on identifying color patches from a user glove. just wondering how i’ll be able to select these color patches if they are located differently for every frame. thanks