I need to find standard deviation from labeled matrix

My project need to find homogeneity from each cell from image so I decided to use SD for each cell

but I already try std(A) but it doesn't work I try meanintensity in regionprops but I don't know what to do next

do I have anyway to get SD from each labelmatrix

clear
clc
close all

IMG = imread('Cal-IR-MP-3.jpg');
blue = IMG(:,:,3);
th = graythresh(blue);
bw = imbinarize(blue,th);
bw = imclearborder(bw,8);


seD = strel('disk',3);
bw = imopen(bw,seD);
CC = bwconncomp(bw, 8);
L = labelmatrix(CC);
xx = [];
a = 1;
LL1 = L;
for i = 1:CC.NumObjects
    LL = L;
    if length(CC.PixelIdxList{i})<1000
        LL(LL==i) = 0;
        LL1(LL1==i) = 0;
    else
        LL(LL ~= i) = 0;
        LL(LL == i)= 1;
        xx(a,1) = i;
        stats = regionprops(LL,blue,'MeanIntensity');
        xx(a,2) = stats.MeanIntensity;
        s = std2(i);
        a = a+1;
    end

    
end

Answer 1

[DGM]

If what you're trying to do is get the standard deviation of the regions specified by the label matrix, try this.

clear variables; clc; close all

IMG = imread('sources/blacklight2.jpg'); % i used my own test image
blue = IMG(:,:,3);
th = graythresh(blue);
bw = imbinarize(blue,th);
bw = imclearborder(bw,8);

seD = strel('disk',3);
bw = imopen(bw,seD);
CC = bwconncomp(bw, 8);
L = labelmatrix(CC);

xx = [];
a = 1;
LL1 = L;
for i = 1:CC.NumObjects
    % only evaluate groups above a certain size
    if length(CC.PixelIdxList{i})<1000 
        % idk what you're doing with LL1
		% i'll assume this is for something else
		% LL was being rewritten and unused, so I removed it
        LL1(LL1==i) = 0;
	else
		roidata=double(blue(L==i)); % extract the ROI
        xx(a,:) = [i std(roidata(:))]; % write to output array
        a = a+1;
    end
end

xx

If LL1 isn't used for anything outside this scope, then it simplifies to

clear variables; clc; close all

IMG = imread('sources/blacklight2.jpg');
blue = IMG(:,:,3);
th = graythresh(blue);
bw = imbinarize(blue,th);
bw = imclearborder(bw,8);

seD = strel('disk',3);
bw = imopen(bw,seD);
CC = bwconncomp(bw, 8);
L = labelmatrix(CC);

xx = [];
a = 1;
for i = 1:CC.NumObjects
    if length(CC.PixelIdxList{i})>=1000
        roidata=double(blue(L==i));
        xx(a,:) = [i std(roidata(:))];
        a = a+1;
	end  
end

xx

idk if that helps

[Image Analyst]

You can use bwareafilt() or bwareaopen() in advance of the labeling to avoid having to check blobs less than 1000 pixels inside the loop.

[DGM]

I was kind of assuming that maybe those small groups might be used for something else. Otherwise, yeah. That would clean things up a bit and potentially save some time as well.

IMG = imread('sources/blacklight2.jpg');
blue = IMG(:,:,3);
th = graythresh(blue);
bw = imbinarize(blue,th);
bw = imclearborder(bw,8);

seD = strel('disk',3);
bw = imopen(bw,seD); % idk that this is needed if using bwareaopen
bw = bwareaopen(bw,1000,8);
CC = bwconncomp(bw, 8);
L = labelmatrix(CC);

xx = zeros([CC.NumObjects 2]);
for i = 1:CC.NumObjects
	roidata=double(blue(L==i));
	xx(i,:) = [i std(roidata(:))];
end