Select image region using point coordinates

Hi, I was wondering if someone could do quickly and efficiently the selection of a region within the multiple ones that matlab manages to detect using the regionprops function in whose interior is a point that is defined by its x and y coordinates.

Answer 1

[Image Analyst]

The built-in function for this is bwselect(). Like

oneBlob = bwselect(multiBlobMask, x, y); % Return only the single blob that contains (x, y).

Here is the help:

bwselect

Select objects in binary image

Syntax

BW2 = bwselect(BW,c,r)

BW2 = bwselect(BW,c,r,n)

[BW2,idx] = bwselect(___)

[x,y,BW2,idx,xi,yi] = bwselect(___)

[___] = bwselect(x,y,BW,xi,yi,n)

[___] = bwselect(BW,n)

[___] = bwselect

Description

BW2 = bwselect(BW,c,r) returns a binary image containing the objects that overlap the pixel (r, c). Objects are connected sets of on pixels, that is, pixels having a value of 1.

[Alejandro Fernández]

Wow thank you so much! I had never seen that feature built in before.

[Image Analyst]

Alejandro, there is another related function that you maybe interested in. It's called imreconstruct(). With that function you give it a mask (with multiple blobs in it), and a "marker" image that has a subset of blobs. Then when you call imreconstruct, it will return only those blobs that overlap any part of a marker blob. So instead of giving it a coordinate or list of coordinates to tell it which blob(s) to extract, you give it an image.

So for example let's say you had an aerial photo of your neighborhood and wanted to find lawns that had white flowering trees on them (like many of them in my neighborhood do this week). So you'd have one mask of only the lawns, then another "marker" image of only the trees, then after you call imreconstruct, you'd get only those lawns that had the white flowering trees on them, and not any of the other lawns.

Answer 2

[Matt J]

regions=regionprops(yourImage,'PixelList');
detector=@(s) ismember([x,y], s.PixelList,'rows');

find( arrayfun( detector   ,   region  ) )

[Alejandro Fernández]

thank you so much for your help. However I think that the option offered by image Analist is much more compact and I have to give the acceptance to him.