How to detecct specific circle and set values inside to 0

Hey everyone,

I have a set of ultra-sound images. Each image have a circle which can be in different positions (this circle is actually the transducer which take those images).

I would like to remove non-zero pixels inside those circles (set the whole circle area to 0). I've tried to use imfindcircles but the radius range of the circles is dynamic.

How can I find those circles and set the whole area inside them to 0?

here are some examples to the images.

Answer 1

[DGM]

There are probably other ways to do this, but this is what I rolled:

inpict = rgb2gray(imread('us2.bmp'));

% generate a mask
bb = inpict>5; % some threshold
bb = despeckle(bb,100); % remove small spots/holes
bb = bwpropfilt(~bb,'solidity',1); % find most solid circle

% apply the mask
inpict(bb) = 0;

If you need the mask to trim off some of the very inner content, you can dilate it after the bwpropfilt() call. This example uses despeckle() from MIMT, but you can do the same with bwareaopen()

bb = ~bwareaopen(~bwareaopen(bb,100),100);

[Itzhak Mamistvalov]

hey, thanks for your answer but i cant find any despeckle function.

is there any other function which performing the same operation?

[DGM]

As I mentioned, despeckle is part of MIMT. You'd have to download it.

Otherwise, you can use bwareaopen() as described.

[Itzhak Mamistvalov]

Thanks! works wonderful