I = imread("barcodeQR.jpg");

% Search the image for a QR Code.
[msg, ~, loc] = readBarcode(I);

% Annotate the image with the decoded message.
xyText =  loc(2,:);
Imsg = insertText(I, xyText, msg, "BoxOpacity", 1, "FontSize", 25);

% Insert filled circles at the finder pattern locations.
Imsg = insertShape(Imsg, "FilledCircle", [loc, ...
    repmat(10, length(loc), 1)], "Color", "red", "Opacity", 1);

% Display image.
imshow(Imsg)

I = imread("barcode1D.jpg");

% Read the 1-D barcode and determine the format..
[msg, format, locs] = readBarcode(I);

% Display the detected message and format.
disp("Detected format and message: " + format + ", " + msg)

% Insert a line to show the scan row of the barcode.
xyBegin = locs(1,:); imSize = size(I);
I = insertShape(I,"Line",[1 xyBegin(2) imSize(2) xyBegin(2)], ...
    "LineWidth", 7);

% Insert markers at the end locations of the barcode.
I = insertShape(I, "FilledCircle", [locs, ...
    repmat(10, length(locs), 1)], "Color", "red", "Opacity", 1);

% Display image.
imshow(I)

I = imread("rotated1DBarcode.jpg");

% Display the image.
imshow(I)

% Pass the image to the readBarcode function.
readBarcode(I)

% Rotate the image by 30 degrees clockwise.
Irot = imrotate(I, -30);

% Display the rotated image.
imshow(Irot)

% Pass the rotated image to the readBarcode function.
readBarcode(Irot)

I = imread("multiple1DBarcodes.jpg");

% ROIs obtained using drawrectangle
roi = [180 100 330 180
    180 320 330 180
    180 550 330 180];

imSize = size(I);
for i = 1:size(roi,1)
    [msg, format, locs] = readBarcode(I, roi(i,:));
    disp("Decoded format and message: " + format + ", " + msg)
    
    % Insert a line to indicate the scan row of the barcode.
    xyBegin = locs(1,:);
    I = insertShape(I,"Line",[1 xyBegin(2) imSize(2) xyBegin(2)], ...
        "LineWidth", 5);
    
    % Annotate image with decoded message.
    I = insertText(I, xyBegin, msg, "BoxOpacity", 1, "FontSize", 20);
end

imshow(I)

I = imread("multiple1DBarcodesRotated.jpg");
Igray = rgb2gray(I);

% Display the image.
imshow(I)

% Pass the unprocessed image to the readBarcode function.
readBarcode(Igray, '1D')

% Detect MSER features.
[~, cc] = detectMSERFeatures(Igray);

% Compute region properties MajorAxisLength and MinorAxisLength.
regionStatistics = regionprops(cc, 'MajorAxisLength', 'MinorAxisLength');

% Filter out components that have a low aspect ratio as unsuitable
% candidates for the bars in the barcode.
minAspectRatio = 10;
candidateRegions = find(([regionStatistics.MajorAxisLength]./[regionStatistics.MinorAxisLength]) > minAspectRatio);

% Binary image to store the filtered components.
BW = false(size(Igray));

% Update the binary image.
for i = 1:length(candidateRegions)
    BW(cc.PixelIdxList{candidateRegions(i)}) = true;
end

% Display the binary image with the filtered components.
imshow(BW)
title("Candidate regions for the barcodes")

% Perform hough transform.
BW = edge(BW,'canny');
[H,T,R] = hough(BW);

% Display the result of the edge detection operation.
imshow(BW)

% Determine the size of the suppression neighborhood.
reductionRatio = 500;
nhSize = floor(size(H)/reductionRatio);
idx = mod(nhSize,2) < 1;
nhSize(idx) = nhSize(idx) + 1;

% Identify the peaks in the Hough transform.
P  = houghpeaks(H,length(candidateRegions),'NHoodSize',nhSize);

% Detect the lines based on the detected peaks.
lines = houghlines(BW,T,R,P);

% Display the lines detected using the houghlines function.
Ihoughlines = ones(size(BW));

% Start and end points of the detected lines.
startPts = reshape([lines(:).point1], 2, length(lines))';
endPts = reshape([lines(:).point2], 2, length(lines))';

Ihoughlines = insertShape(Ihoughlines, 'Line', [startPts, endPts], ...
    'LineWidth', 2, 'Color', 'green');

% Display the original image overlayed with the detected lines.
Ibarlines = imoverlay(I, ~Ihoughlines(:,:,1));
imshow(Ibarlines)

useClustering = license('test','statistics_toolbox');

if useClustering
    [boundingBox, orientation, Iclusters] = clusteringLocalization(lines, size(I));
    
    % Display the detected clusters.
    imshow(Iclusters)
else
    disp("The clustering based workflow requires a license for the Statistics and Machine Learning Toolbox")
end

if ~useClustering
    [boundingBox, orientation, Idilated] = segmentationLocalization(Ihoughlines);
    
    % Display the dilated image.
    imshow(Idilated)
end

% Localize and rotate the barcodes in the image.
correctedImages = cell(1, length(orientation));

% Store the cropped and rotation corrected images of the barcodes.
for i = 1:length(orientation)
    
    I = insertShape(I, 'Rectangle', boundingBox(i,:), 'LineWidth',3, 'Color', 'red');
    
    if orientation(i) > 0
        orientation(i) = -(90 - orientation(i));
    else
        orientation(i) = 90 + orientation(i);
    end
    
    % Crop the barcode from the original image and rotate it using the
    % detected orientation.
    correctedImages{i} = imrotate(imcrop(Igray,boundingBox(i,:)), orientation(i));
end

% Display the image with the localized barcodes.
imshow(I)

% Pass each of the images to the readBarcode function.
for i = 1:length(correctedImages)
    [msg, format, ~] = readBarcode(correctedImages{i}, '1D');
    disp("Decoded format and message: " + format + ", " + msg)
end

function [boundingBox, orientation, Iclusters] = clusteringLocalization(lines, imSize)

%------------------------------------------------------------------------
% Determine Bisectors of Barcode Line Segments
%------------------------------------------------------------------------

% Table to store the properties of the bisectors of the detected lines.
linesBisector = array2table(zeros(length(lines), 4), 'VariableNames', {'theta', 'rho', 'x', 'y'});

% Use the orientation values of the lines to determine the orientation.
% values of the bisectors
idxNeg = find([lines.theta] < 0);
idxPos = find([lines.theta] >= 0);

negAngles = 90 + [lines(idxNeg).theta];
linesBisector.theta(idxNeg) = negAngles;

posAngles = [lines(idxPos).theta] - 90;
linesBisector.theta(idxPos) = posAngles;

% Determine the midpoints of the detected lines.
midPts = zeros(length(lines),2);

% Determine the 'rho' values of the bisectors.
for i = 1:length(lines)
    midPts(i,:) = (lines(i).point1 + lines(i).point2)/2;
    linesBisector.rho(i) = abs(midPts(i,2) - tand(lines(i).theta) * midPts(i,1))/...
        ((tand(lines(i).theta)^2 + 1) ^ 0.5);
end

% Update the [x,y] locations of the bisectors using their polar
% coordinates.
[linesBisector.x, linesBisector.y] = pol2cart(deg2rad(linesBisector.theta),linesBisector.rho,'ro');

%------------------------------------------------------------------------
% Perform Clustering on the Bisectors to Identity the Individual Barcodes
%------------------------------------------------------------------------

% Store the [x,y] data of the bisectors to be used for clustering.
X = [linesBisector.x,linesBisector.y];

% Get pairwise distance between the points
D = pdist2(X,X);

% Perform density-based spatial clustering to separate the different
% barcodes in the image.
searchRadius = max(imSize/5);
minPoints = 10;
idx = dbscan(D,searchRadius, minPoints);

% Identify the number of clusters (barcodes).
numClusters = unique(idx(idx > 0));

% Store the endpoints of the detected lines.
dataXY = cell(1, length(numClusters));

% Image to show the detected clusters (barcodes).
Iclusters = ones(imSize);

for i = 1:length(numClusters)
    classIdx = find(idx == i);
    
    rgbColor = rand(1,3);
    startPts = reshape([lines(classIdx).point1], 2, length(classIdx))';
    endPts = reshape([lines(classIdx).point2], 2, length(classIdx))';
    
    % Insert lines corresponding to the current cluster (barcode).
    Iclusters = insertShape(Iclusters, 'Line', [startPts, endPts], ...
        'LineWidth', 2, 'Color', rgbColor);
    
    % Update the endpoints of the lines in each cluster (barcode).
    dataXY{i} = [startPts; endPts];
end

%------------------------------------------------------------------------
% Localization parameters for the barcode
%------------------------------------------------------------------------

orientation = zeros(1,length(numClusters));
boundingBox = zeros(length(numClusters), 4);

% Padding the cropped images of barcodes.
padding = 40;

% Determine the ROI and orientation of the individual clusters (barcodes).
for i = 1:length(numClusters)
    
    % Bounding box coordinates with padding.
    x1 = min(dataXY{i}(:,1)) - padding;
    x2 = max(dataXY{i}(:,1)) + padding;
    y1 = min(dataXY{i}(:,2)) - padding;
    y2 = max(dataXY{i}(:,2)) + padding;
    
    boundingBox(i,:) = [x1, y1, x2-x1, y2-y1];
    
    % Orientation of the barcode.
    orientation(i) = mean(linesBisector.theta(idx == i));
    
end

end

function [boundingBox, orientation, Idilated] = segmentationLocalization(Ihoughlines)

%------------------------------------------------------------------------
% Use image dilation to separate the barcodes
%------------------------------------------------------------------------

% Create binary image with the detected lines.
Ibw = ~Ihoughlines(:,:,1);
Ibw(Ibw > 0) = true;

% Dilate the image using a disk structuring element.
diskRadius = 10; % Might need tuning depending on the input image.
se = strel('disk', diskRadius);
Idilated = imdilate(Ibw, se);

%------------------------------------------------------------------------
% Localization parameters for the barcode
%------------------------------------------------------------------------

% Compute region properties Orientation and BoundingBox.
regionStatistics = regionprops(Idilated, 'Orientation', 'BoundingBox');

% Padding for the cropped images of barcodes.
padding = 40;

boundingBox = zeros(length(regionStatistics), 4);

for idx = 1:length(regionStatistics)
    
    boundingBox(idx,:) = regionStatistics(idx).BoundingBox;
    
    % Bounding box coordinates with padding.
    boundingBox(idx,1) = boundingBox(idx,1) - padding;
    boundingBox(idx,2) = boundingBox(idx,2) - padding;
    boundingBox(idx,3) = boundingBox(idx,3) + 2*padding;
    boundingBox(idx,4) = boundingBox(idx,4) + 2*padding;
    
end

orientation = [regionStatistics(:).Orientation];

end