can someone help me edit the code below so that the output image can display the value of rotation angle of the object with axis guide

It is just the example image that i want, the axis with the angle:

this is my input image:

% Perform edge detection with interpolation during non maximum suppression
function CannyEdgeDetector()
    close all;  % Close figures
    saveImage = true;
    sigma = 1; % Gaussian filter sigma
    highThresholdRatio = 0.275; % High threshold ratio
    lowThresholdRatio = 0.25; % Low threshold ratio OF THE high threshold
    
    % Change the current folder to the folder of this m-file.
    % Courtesy of Brett Shoelson
    if(~isdeployed)
      cd(fileparts(which(mfilename)));
    end
    
    im = imread('C:\Users\User\Pictures\bt.jpeg');
    figure; imshow(im);
    title('Original Image');
    %if saveImage
        %imwrite(im, 'C:\Users\User\Pictures\O3.jpeg');
    %end
    
    % Smooth with Gaussian 5x5 filter to reduce noise
    im = rgb2gray(im);
    figure; imshow(im);
    title('B/W Image');
    %if saveImage
        %imwrite(im, 'Output_Photos\2_bw.jpg');
    %end
    
    im = double(imgaussfilt(im,sigma));
    figure; imshow(NormalizeMatrix(im));
    title('Gaussian Filter');
    %if saveImage
        %imwrite(NormalizeMatrix(im), 'Output_Photos\3_gaussian.jpg');
    %end
    
    % Find the intensity gradient of the image
    Gx = SobelFilter(im, 'x');
    Gy = SobelFilter(im, 'y');
    Gx = imgaussfilt(Gx,sigma);
    Gy = imgaussfilt(Gy,sigma);
    figure; imshow(abs(normalize(Gx)));
    title('Gx Sobel Filter');
    %if saveImage
        %imwrite(abs(NormalizeMatrix(Gx)), 'Output_Photos\4_gx_sobel.jpg');
    %end
    figure; imshow(abs(NormalizeMatrix(Gy)));
    title('Gy Sobel Filter');
    %if saveImage
        %imwrite(abs(NormalizeMatrix(Gy)), 'Output_Photos\5_gy_sobel.jpg');
    %end
    
    % Find the magnitude of the gradient
    Gmag = sqrt(Gx.^2 + Gy.^2);
    angle = atan2(Gy,Gx)*180/pi;
    figure; imshow(NormalizeMatrix(Gmag));
    title('Gmag');
    %if saveImage
        %imwrite(NormalizeMatrix(Gmag), 'Output_Photos\6_gmag.jpg');
    %end
         
    % Perform non-maximum suppression using interpolation
    [h,w] = size(im);
    X=[-1,0,+1 ;-1,0,+1 ;-1,0,+1];
	Y=[-1,-1,-1 ;0,0,0 ;+1,+1,+1];
    output = zeros(h,w);
    x = [0 1];
    for i=2:h-1 % row
        for j=2:w-1 % col
            if (angle(i,j)>=0 && angle(i,j)<=45) || ...
                    (angle(i,j)<-135 && angle(i,j)>=-180)
                yBot = [Gmag(i,j+1) Gmag(i+1,j+1)];
                yTop = [Gmag(i,j-1) Gmag(i-1,j-1)];
                x_est = abs(Gy(i,j)/Gmag(i,j)); % y
                if (Gmag(i,j) >= ((yBot(2)-yBot(1))*x_est+yBot(1)) && ...
                    Gmag(i,j) >= ((yTop(2)-yTop(1))*x_est+yTop(1))) % interpolation
                    output(i,j)= Gmag(i,j);
                else
                    output(i,j)=0;
                end
            elseif (angle(i,j)>45 && angle(i,j)<=90) || ...
                    (angle(i,j)<-90 && angle(i,j)>=-135)
                yBot = [Gmag(i+1,j) Gmag(i+1,j+1)];
                yTop = [Gmag(i-1,j) Gmag(i-1,j-1)];
                x_est = abs(Gx(i,j)/Gmag(i,j));
                if (Gmag(i,j) >= ((yBot(2)-yBot(1))*x_est+yBot(1)) && ...
                    Gmag(i,j) >= ((yTop(2)-yTop(1))*x_est+yTop(1)))
                    output(i,j)= Gmag(i,j);
                else
                    output(i,j)=0;
                end
            elseif (angle(i,j)>90 && angle(i,j)<=135) || ...
                    (angle(i,j)<-45 && angle(i,j)>=-90)
                yBot = [Gmag(i+1,j) Gmag(i+1,j-1)];
                yTop = [Gmag(i-1,j) Gmag(i-1,j+1)];
                x_est = abs(Gx(i,j)/Gmag(i,j));
                if (Gmag(i,j) >= ((yBot(2)-yBot(1))*x_est+yBot(1)) && ...
                    Gmag(i,j) >= ((yTop(2)-yTop(1))*x_est+yTop(1)))
                    output(i,j)= Gmag(i,j);
                else
                    output(i,j)=0;
                end
            elseif (angle(i,j)>135 && angle(i,j)<=180) || ...
                    (angle(i,j)<0 && angle(i,j)>=-45)
                yBot = [Gmag(i,j-1) Gmag(i+1,j-1)];
                yTop = [Gmag(i,j+1) Gmag(i-1,j+1)];
                x_est = abs(Gx(i,j)/Gmag(i,j));
                if (Gmag(i,j) >= ((yBot(2)-yBot(1))*x_est+yBot(1)) && ...
                    Gmag(i,j) >= ((yTop(2)-yTop(1))*x_est+yTop(1)))
                    output(i,j)= Gmag(i,j);
                else
                    output(i,j)=0;
                end
            end           
        end
    end
    
    Gmag = NormalizeMatrix(output);
    figure; imshow(Gmag);
    title('Non Maximum Suppression');
    %if saveImage
        %imwrite(Gmag, 'Output_Photos\7_non_maximum_suppression.jpg');
    %end
    
    % Perform double thresholding
    highThreshold = max(max(Gmag))*highThresholdRatio;
    lowThreshold = highThreshold*lowThresholdRatio;
    strongEdgesRow = zeros(1,h*w); % Keep track of the strong edge row index
    strongEdgesCol = zeros(1,h*w); % Keep track of the strong edge col index
    weakEdgesRow = zeros(1,h*w);  % Keep track of the weak edge row index
    weakEdgesCol = zeros(1,h*w);  % Keep track of the weak edge col index
    strongIndex = 1;
    weakIndex = 1;
    for i=2:h-1 % row
        for j=2:w-1 % col
            if Gmag(i,j) > highThreshold    % Strong edge
                Gmag(i,j) = 1;
                strongEdgesRow(strongIndex) = i;
                strongEdgesCol(strongIndex) = j;
                strongIndex = strongIndex + 1;
            elseif Gmag(i,j) < lowThreshold % No edge
                Gmag(i,j) = 0;
            else                            % Weak edge
                weakEdgesRow(weakIndex) = i;
                weakEdgesCol(weakIndex) = j;
                weakIndex = weakIndex + 1;
            end
        end
    end
    figure; imshow(Gmag);
    title('Double Threshold'); 
    %if saveImage
        %imwrite(Gmag, 'Output_Photos\8_double_threshold.jpg');
    %end
    
    % Perform edge tracking by hysteresis
    set(0,'RecursionLimit',10000)
    for i=1:strongIndex-1
        % Find the weak edges that are connected to strong edges and set 
        % them to 1
        Gmag = FindConnectedWeakEdges(Gmag, strongEdgesRow(i),...
            strongEdgesCol(i));
    end
    figure; imshow(Gmag);
    title('Edge Tracking Before Clean Up'); 
    %if saveImage
        %imwrite(Gmag, 'Output_Photos\9_edge_tracking.jpg');
    %end
    
    % Remove the remaining weak edges that are not actually edges
    % and is noise instead
    for i=1:weakIndex-1
        if Gmag(weakEdgesRow(i),weakEdgesCol(i)) ~= 1
            Gmag(weakEdgesRow(i),weakEdgesCol(i)) = 0;
        end
    end
    figure; imshow(Gmag);
    title('Edge Tracking After Clean Up'); 
    %if saveImage
        %imwrite(Gmag, 'Output_Photos\10_final.jpg');
    %end
    
%     % MATLAB canny comparison
%     im = imread('Test_Photos/test1.jpg');
%     im = rgb2gray(im);
%     im = edge(im, 'canny');
%     figure; imshow(im);
%     title('MATLAB');
end

% Normalize matrix
function[A] = NormalizeMatrix(A)
    A = A/max(A(:));
end


% Perform sobel filter
function[A] = SobelFilter(A, filterDirection)
    switch filterDirection
        case 'x' 
            Gx = [-1 0 +1; -2 0 +2; -1 0 +1];
            A = imfilter(A, double(Gx), 'conv', 'replicate');
        case 'y'
            Gy = [-1 -2 -1; 0 0 0; +1 +2 +1];
            A = imfilter(A, double(Gy), 'conv', 'replicate');
        otherwise
            error('Bad filter direction - try inputs ''x'' or ''y''');
    end
end

% Find weak edges that are connected to strong edges and set them to 1
function[Gmag] = FindConnectedWeakEdges(Gmag, row, col)
    for i = -3:1:3
        for j = -3:1:3
            if (row+i > 0) && (col+j > 0) && (row+i < size(Gmag,1)) && ...
                    (col+j < size(Gmag,2)) % Make sure we are not out of bounds
                if (Gmag(row+i,col+j) > 0) && (Gmag(row+i,col+j) < 1)
                    Gmag(row+i,col+j) = 1;
                    Gmag = FindConnectedWeakEdges(Gmag, row+i, col+j);
                end
            end
        end
    end
end

Answer 1

[Image Analyst]

I'd use bwferet and get the min angle. Let me know if you can't figure it out.

[Nurul Farhana Mohd Fadzli]

Can you help me using the code i attached above?

[DGM]

https://www.mathworks.com/help/images/ref/regionprops.html#buoixjn-1-properties

https://www.mathworks.com/help/images/ref/bwferet.html

A = imread('toothlotion.png');
A = rgb2gray(A)>128;

% get properties
S = regionprops(A,'minferetproperties','centroid');
th = vertcat(S.MinFeretAngle);
C = vertcat(S.Centroid); % just used for placing labels

% reshape coordinates for plotting diameters
mfcoords = vertcat(S.MinFeretCoordinates);
x = reshape(mfcoords(:,1),2,[]).';
y = reshape(mfcoords(:,2),2,[]).';

% plot everything
imshow(A); hold on
for k = 1:size(x,1)
    plot(x(k,:),y(k,:),'linewidth',3);
    
    t = text(C(k,1),C(k,2),num2str(round(th(k))));
    t.Color = 'r';
    t.HorizontalAlignment = 'center';
end

[Nurul Farhana Mohd Fadzli]

How about when there is only one object?

[DGM]

Same thing. Instead of having 5 angles, you'll just have one angle.

[Nurul Farhana Mohd Fadzli]

Thank you,so i just can simply use the code? im sorry im still new to matlab. but why i get output like this

[DGM]

Probably because you need to get a clean binary image with only one object in it. You may be able to use bwareafilt() to do so, though it may take some other work to get the object edges cleaned up. It's hard to tell.

[Nurul Farhana Mohd Fadzli]

okayy thank you for your help sir

[Nurul Farhana Mohd Fadzli]

@DGM Sir, I want to deploy my coding to Nvidia Jetson but I just found out that minFeretProperties function are not supported for code generation in MATLAB Coder. Can you help me with another function that i can use to find the angle of object that can produce the same output like the above example ?

[DGM]

I'm sorry, but I really have no idea what will work for code generation. I don't think I even have those tools.

[Nurul Farhana Mohd Fadzli]

@DGMIts okay sir you help me a lot. Sir can you show me how to find the angle using regionprops(__,'Orientation'). I have tried this simple code below and the output angle is sometimes not correct but I will try. And I wish you could also help me producing an output image with an angle display on it like your example above.

[Image Analyst]

regionprops looks at the angle of an ellipse fitted to the blob boundary. It's not always accurate for other shapes, especially rectangles. That's why I suggested bwferet instead of regionprops.

[Nurul Farhana Mohd Fadzli]

I already use that, like DGM's example above which is the minFeretProperties but the problem is, this function is not supported for code generation. bwferet also not supported for code generation. Now im trying to edit DGM's example above using other function that can also detect the angle of object. Can you help me sir. @Image Analyst

[Image Analyst]

Then I suggest you find the algorithm for feret diameter and program up your own version. I can't help you with that.

[DGM]

Like @Image Analyst says (and I think I saw in your other thread), using "orientation" tends to orient rectangles along their diagonals, so I really doubt that's what you want.

regionprops(..., 'minferetproperties') relies on bwferet(). Not knowing what works with coder, I'm not sure what part of bwferet() is incompatible. It seems to all be straightforward, but maybe convhull() is the problem.

I don't have another implementation of bwferet().

[Nurul Farhana Mohd Fadzli]

Did you know how to find angle without calling function? using formula maybe?

[DGM]

If you open up bwferet(), you can basically see all the math involved in finding the feret properties. You might be able to make some simplifications if you could rely on your objects always being rectangles, or at least always being nominally convex.

Answer 2

[yanqi liu]

yes，sir，may be use minboundrect to get the rect area，such as

[Nurul Farhana Mohd Fadzli]

Thank you but can i know how to get the rotation angle of the box ? It will be good if its also displaying the axis

[Nurul Farhana Mohd Fadzli]

Nevermind ,how about creating bounding box that exactly foloow the shape of the box? Please, I really need help