Why is Matlab detecting only one color

Question

Jonathan 2019-10-23

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/486982-why-is-matlab-detecting-only-one-color

评论： Guillaume 2019-10-23

I am trying to detect two colors using matlab, red and blue. I want to detect either a red or blue object so that the robot will come and pick the object and place it in the appropriate bin according to its color. However, the code I wrote is only detecting one color. In place of "for" I tried to use "if" and "elseif" statement but still it was detecting only one of the two colors. What could be the problem? Shown below is my code:

%Initialization
redThresh = 0.24; % Threshold for red detection
blueThresh = 0.15; % Threshold for blue detection
vid = videoinput('winvideo',1);
set(vid, 'FramesPerTrigger', Inf);
set(vid, 'ReturnedColorspace', 'rgb')
vid.FrameGrabInterval = 10;
nFrame = 0; % Frame number initialization
start(vid)
% Processing loop
while(nFrame < 10)
    s = getsnapshot(vid); % Acquire single frame
     for redThresh=0.24 %&& redThresh <= 0.26
        % s=rgbFrame();
        red = s(:,:,1);
        diffFrameRed = imsubtract(s(:,:,1), rgb2gray(s));
        diffFrameRed = medfilt2(diffFrameRed, [3 3]);
        binFrameRed = imbinarize(diffFrameRed, redThresh);
        binFrameRed = bwareaopen(binFrameRed,300);
        bw = bwlabel(binFrameRed, 8);
        
        stats = regionprops((bw), 'BoundingBox', 'Centroid');
        centroids = cat(1, stats.Centroid);
        imshow(s);
        hold on
        for object = 1:length(stats)
            bb = stats(object).BoundingBox;
            bc = stats(object).Centroid;
            rectangle('Position',bb,'EdgeColor','r','LineWidth',2)
            plot(bc(1),bc(2), '-m+')
            a=text(bc(1)+15,bc(2), strcat('X: ', num2str(round(bc(1))), '    Y: ', num2str(round(bc(2)))));
            set(a, 'FontName', 'Arial', 'FontWeight', 'bold', 'FontSize', 12, 'Color', 'yellow');
            X = bc(1); Y = bc(2);
        end
        hold off
    end
    for blueThresh = 0.13% && blueThresh <= 0.16
        blue = s(:,:,3);
        diffFrameBlue = imsubtract(s(:,:,3), rgb2gray(s));
        diffFrameBlue = medfilt2(diffFrameBlue, [3 3]);
        binFrameBlue = imbinarize(diffFrameBlue, blueThresh);
        binFrameBlue = bwareaopen(binFrameBlue,300);
        bw = bwlabel(binFrameBlue, 8);
        
        stats = regionprops((bw), 'BoundingBox', 'Centroid');
        centroids = cat(1, stats.Centroid);
        imshow(s);
        hold on
        for object = 1:length(stats)
            bb = stats(object).BoundingBox;
            bc = stats(object).Centroid;
            rectangle('Position',bb,'EdgeColor','b','LineWidth',2)
            plot(bc(1),bc(2), '-m+')
            a=text(bc(1)+15,bc(2), strcat('X: ', num2str(round(bc(1))), '    Y: ', num2str(round(bc(2)))));
            set(a, 'FontName', 'Arial', 'FontWeight', 'bold', 'FontSize', 12, 'Color', 'blue');
            X = bc(1); Y = bc(2);
        end
        hold off
    end   
    pause(5)
    nFrame = nFrame+1;
end

4 个评论
显示 2更早的评论隐藏 2更早的评论

Jonathan 2019-10-23

@Rik and @Guillaume

Thank you so much for your responses.

I want that when the colored components are moving on a conveyor, when the red one is in the field of view then it is detected, then picked and placed to a respective location. Then when a blue one is in view, in the same way, it is detected and then picked by the robot to a respective location. How am I supposed to code for that to happen?

Your help is greatly appreciated

Guillaume 2019-10-23

在 MATLAB Online 中打开

It seems to me that if you remove the two lines

for redThresh=0.24 %&& redThresh <= 0.26
for blueThresh = 0.13% && blueThresh <= 0.16

your code should work as designed. First detect the red objects and highlight them, then detect the blue objects and highlight them. If an object is detected as both red and blue then it'd be highlighted as blue since that the last thing done.

I assume you've read up on colour detection and selected your algorithm accordingly. Your algorithm won't work on generic images but perhaps for your use case, 'red' and 'blue' are sufficiently separated in the RGB colour space.

This is the result of your algorithm on an artificial image: