How do I find the centre of an image?

I want to find the centre of the white area and measure the length of the x and y axes.

[Sarvesh Kale]

It would be nice if you upload the image as an attachment

[DGM]

As I have a feeling that the question being asked isn't what's actually needed, see also:

https://www.mathworks.com/matlabcentral/answers/1906470-how-do-i-measure-distances-in-an-image

Jonas' answer gives the section widths intersecting the blob centroid -- what the question asks for.

The above link also covers similar size metrics:

• height and width of bounding box
• equivalent diameters
• feret diameters

Sarvesh's answer below already touched on two of those.

Answer 1

[Jonas]

you can use regionprops for that and let the function return the centroid. This may be the cetner you are searching for.

If you are then interested in the horizontal and vertical extent trough this point, you could just search for the first/last white pixel in the row/column given by the coordinates of the centroid. given centroidX,centroidY you could use something like

find(img(:,centroidX),1,'last')-find(img(:,centroidX),1,'last')
% and
find(img(centroidY,:),1,'last')-find(img(centroidY,:),1,'last')

[Kantachai Chamnarnkit]

I don't know how to search for the first/last white pixel in the row/column

[DGM]

You use find() like Jonas mentioned.

img = imread('tiltellipse.png')>128;

% get the centroid
S = regionprops(img,'centroid');
centroidX = round(S.Centroid(1)); % round it for indexing purposes
centroidY = round(S.Centroid(2));

% i'm doing it this way so that i can plot the results
yrange = [find(img(:,centroidX),1,'first') find(img(:,centroidX),1,'last')];
xrange = [find(img(centroidY,:),1,'first') find(img(centroidY,:),1,'last')];
blobheight = diff(yrange)

blobheight = 128

blobwidth = diff(xrange)

blobwidth = 121

% plot the lines that are being measured
imshow(img); hold on
plot(xrange,[1 1]*centroidY,'r-o');
plot([1 1]*centroidX,yrange,'b-o');

Note that this gives the maximum width along each line. Consider how this would behave with an object like this:

It's up to you to decide what you want to measure.

Answer 2

[Sarvesh Kale]

Hi Kantachai Chamnarnkit,

the following code snippet might help you. Assuming the image in binary in nature

i = imread('image.jpeg'); % name of your image should be inside ' '
blobs=regionprops(i,"Centroid","MinorAxisLength","MajorAxisLength","BoundingBox");

for more information on regionprops

https://in.mathworks.com/help/images/ref/regionprops.html

[Jonas]

MinorAxisLength / MajorAxisLength does not give horizontal and vertical size of the blob. it assumes an ellipse and takes the axes of the ellipse, which can be rotated

the bounding box has the problem that the respective line will not necessarily go through the center