Asked by Itzik Ben Shabat
on 16 May 2013

Hi, I have a polygon defined by its vertices in 2D (i.e a nX2 matrix). I have a pixel image of size MxM. i wish to label all the pixels that are on the boundary f the polygon. is there a function that does this? i know inpolygon, but it labels all vertices that are inside the polygon, all i need is the boundary. any thoughts? thanks

Answer by Thomas
on 16 May 2013

Edited by Thomas
on 16 May 2013

regionprops might help

doc regionprops

look at the section 'perimeter'

http://www.mathworks.com/help/images/ref/regionprops.html#bqkf8jn

This might help as well

http://angeljohnsy.blogspot.com/2012/01/boundary-extraction-in-matlab.html

Answer by Sean de Wolski
on 16 May 2013

You could use `poly2mask()` to create a mask of your polygon and then use `bwboundaries` or `bwtraceboundaries()` to draw the edges.

This will then give you a logical mask of the edges of your polygon. You can use this to do whatever you want to do, e.g. change the color in the original image etc.

Itzik Ben Shabat
on 16 May 2013

Answer by Matt Kindig
on 16 May 2013

You can use the inpoly() function on the file exchange ( http://www.mathworks.com/matlabcentral/fileexchange/10391-fast-points-in-polygon-test). It can return points on the boundary separately from those inside. It's also a lot faster than the built-in inpolygon().

Answer by Image Analyst
on 16 May 2013

Edited by Image Analyst
on 16 May 2013

Like Sean said, first use poly2mask() to create a binary image of a solid polygon from your list of perimeter coordinates. This solid polygon can be used as a mask.

Then, from the binary image and want the (x,y) coordinates of all of the pixels in the boundary, use bwboundaries() in the Image Processing Toolbox. See my Image Segmentation Tutorial for an example: http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862

If you want just the convex hull points (like where the rubber band would contact your object if you wrapped a rubber band around it) use convhull().

If you want an image of the perimeter pixels, use bwperim() in the Image Processing Toolbox. You can use this as a mask also, if you want.

Show 1 older comment

Itzik Ben Shabat
on 17 May 2013

Image Analyst
on 17 May 2013

Sean de Wolski
on 17 May 2013

Here's how I would approach this Itzik:

Find the minimum x and y values of your coordinates. Add the absolute values of these points + 1 to your x and y coordinates. Essentially this is shifting your points to be strictly greater than or equal to 1.

Then use `poly2mask` and IA/my steps from above. Then subtract off the values you added.

Opportunities for recent engineering grads.

## 0 Comments