Measuring percentage of black pixels

I have four ROIs labelled as a mask. I want to measure the number and percentage of black pixels (black pixels/ total ROI area) in the 'binary' image inside the individual ROIs defined above.

Second question - I have multiple 'binary' images to analyze but the ROIs will be the same. Any suggestions on batch processing?

Answer 1

[Simon Chan]

Try the following:

Variables 'dark_number' and 'dark_percent' reports the number and percentage of black pixels in each ROI respectively.

Each row refers to each binary image and each column refers to each ROI. So the number of columns are always 4 while row number varies

clear; clc;
[filename_roi,pathname_roi] = uigetfile('*.png','Select ROI png Files','MultiSelect', 'on');
if ~iscell(filename_roi)
    filename_roi = {filename_roi};
end
Nroi = length(filename_roi);
ROI = cell(1,Nroi);
for r = 1:Nroi
    ROI{1,r} = imread(fullfile(pathname_roi,filename_roi{r}));
end
%
[filename,pathname] = uigetfile('*.png','Select One or More png Files','MultiSelect', 'on');
if ~iscell(filename)
    filename = {filename};
end
Nz = length(filename);
white_mask = cellfun(@nnz,ROI);
dark_number = zeros(Nz,Nroi);
for k = 1:Nz
    I = imread(fullfile(pathname,filename{k}));
    white_number = cellfun(@(x) sum((I & x),'all'),ROI);
    dark_number(k,:) = white_mask - white_number;
end
dark_percent = 100*dark_number./white_mask

[Aero Descerazes]

@Simon Chan, this works great. I have another question -

Is there a way to print the row names (image filename), and the column names (ROI names) in a consolidated output table? It is diffcult to know the files and regions.

Thanks a lot!

[Simon Chan]

Add the following lines:

Edit: Use the ROI filename as the column names.

colName = extractBefore(filename_roi,'.');
T1 = array2table(dark_number,'VariableNames',colName);
T2 = array2table(dark_percent,'VariableNames',colName);
T = table(T1,T2,'VariableNames',{'Dark Pixel Number inside ROI','Percentage of Dark Pixels inside ROI'},'RowName',filename)

Sample results as follows:

[Aero Descerazes]

@Simon Chan. Thanks for the reply. However, if there are many ROI, it still does not show. Is there a way to convert the table T to a homogeneous matrix?

I am asking because I want to save this as a CSV file.

[Simon Chan]

Actually, both results are already stored in table T1 (Number of Dark Pixel) and T2 (Percentage of Dark Pixel). You can just type T1 and T2 and results as follows. However, if you increase the number of ROIs continously, the same issue will happen again.

Then you have to think about how would you like to present your results. May be function fprintf suggested by yanqi liu is better if you have lots of ROIs.

[Aero Descerazes]

@Simon Chan Thank you very much for the suggestion!

Answer 2

[yanqi liu]

clc; clear all; 
rois = {'https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/884800/ROI%201.png',...
    'https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/884805/ROI%202.png',...
    'https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/884810/ROI%203.png',...
    'https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/884815/ROI%204.png'};
% make all your image filenames
filenames = {'https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/884795/binary.png',...
    'https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/884795/binary.png'};
bws = [];
for i = 1 : length(rois)
    bws{i} = im2bw(imread(rois{i}));
end
for i = 1 : length(filenames)
    imi = im2bw(imread(filenames{i}));
    % mask
    black_pixels = length(find(imi.*bws{i}));
    total_ROI_area = length(find(bws{i}));
    rate = black_pixels/total_ROI_area;
    fprintf('\n%d image---rate=%.2f',i,rate);
end

1 image---rate=0.72
2 image---rate=0.65