How can i find the Majoraxislength and Minoraxislength of each cell present in binary image ?

Answer 1

[Jonas]

a call to regionprops should suffice:

clear;
im=imread("image.png");
im=rgb2gray(im); % to grayscale
im=im~=0; % make binary image
imshow(im)

stats = regionprops('table',im,'Centroid','MajorAxisLength','MinorAxisLength');


% check if the results make sence
centers = stats.Centroid;
diametersMax = max([stats.MajorAxisLength stats.MinorAxisLength],[],2);
diametersMin = min([stats.MajorAxisLength stats.MinorAxisLength],[],2);

radiiMax = diametersMax/2;
radiiMin = diametersMin/2;


hold on
viscircles(centers,radiiMax,'Color','red');
viscircles(centers,radiiMin,'Color','blue');
hold off

[Shri.s]

Big Thanku its working sir......

[Jonas]

maybe it is also better to calculate the convec hull of each object first to make the result a bit cleaner:

clear;
im=imread("image.png");
im=rgb2gray(im); % to grayscale
im=im~=0; % make binary image
im=bwconvhull(im,'objects'); %%%%%%%%%%%% added
imshow(im)

stats = regionprops('table',im,'Centroid','MajorAxisLength','MinorAxisLength');


% check if the results make sence
centers = stats.Centroid;
diametersMax = max([stats.MajorAxisLength stats.MinorAxisLength],[],2);
diametersMin = min([stats.MajorAxisLength stats.MinorAxisLength],[],2);

radiiMax = diametersMax/2;
radiiMin = diametersMin/2;


hold on
viscircles(centers,radiiMax,'Color','red');
viscircles(centers,radiiMin,'Color','blue');
hold off

[Jonas]

maxbe using the min and max feret is a better possibility since the axes sometimes seem not to be correct:

clear;
im=imread("image.png");
im=rgb2gray(im);
im=im~=0;
im=bwconvhull(im,'objects');

imshow(im)

stats = regionprops('table',im,'Centroid','MajorAxisLength','MinorAxisLength','Orientation');


% check if the results are correct
centers = stats.Centroid;
diametersMax = max([stats.MajorAxisLength stats.MinorAxisLength],[],2);
diametersMin = min([stats.MajorAxisLength stats.MinorAxisLength],[],2);

dxMajor=stats.MajorAxisLength.*cosd(stats.Orientation)./2;
dyMajor=stats.MajorAxisLength.*sind(stats.Orientation)./2;

dxMinor=stats.MinorAxisLength.*sind(stats.Orientation)./2;
dyMinor=stats.MinorAxisLength.*cosd(stats.Orientation)./2;

for nr=1:height(stats)
    if stats.Orientation(nr)>0
        xcordsMaj(:,nr)=(centers(nr,1)+[+1 -1]*dxMajor(nr))';
        ycordsMaj(:,nr)=(centers(nr,2)+[-1 +1]*dyMajor(nr))';
        
        xcordsMin(:,nr)=(centers(nr,1)+[+1 -1]*dxMinor(nr))';
        ycordsMin(:,nr)=(centers(nr,2)+[+1 -1]*dyMinor(nr))';
    else
        xcordsMaj(:,nr)=(centers(nr,1)+[+1 -1]*dxMajor(nr))';
        ycordsMaj(:,nr)=(centers(nr,2)+[-1 +1]*dyMajor(nr))';
        
        xcordsMin(:,nr)=(centers(nr,1)+[+1 -1]*dxMinor(nr))';
        ycordsMin(:,nr)=(centers(nr,2)+[-1 +1]*dyMinor(nr))';
    end

end

line([xcordsMaj xcordsMaj ],[ycordsMaj ycordsMin],'Color','red')

the feret looks like that:

clear;
im=imread("image.png");
im=rgb2gray(im);
im=im~=0;
im=bwconvhull(im,'objects'); % not needed here because ferets are calculated over conv hull; neverthless here for better line visibility

imshow(im)

stats = regionprops('table',im,'MaxFeretProperties','MinFeretProperties');

cellfun(@(in)line(in(:,1),in(:,2),'Color','red'),[stats.MinFeretCoordinates; stats.MaxFeretCoordinates])

[Shri.s]

okkk sir thanku....