MATLAB: How to detect or identify the middle row of the crops by hough()

image processingImage Processing Toolboximage segmentation

Best Answer

I don't think you need hough. You can simply use regionprops() and ask for centroid and orientation. You can then use ismember on the labeled image to extract either

the blob with the most vertical orientation, OR
the blob with the x centroid closest to the middle of the image.

Hopefully those are the same blob - I think they will be for the example you posted. Starting with your segmented, binary image, it goes something like this:

labeledImage = bwlabel(binaryImage);
props = regionprops(labeledImage, 'Centroid', 'Orientation');
% Find blob based on angle
diffAngles = abs([props.Orientation] - 90);
[minAngleDiff, index] = min(diffAngles);
centerBlob = ismember(labeledImage, index);
% Find blob based on centroid
[rows, columns] = size(binaryImage)
xyCentroids = vertcat(props.Centroid);
xCentroids = xyCentroids(:, 1);
diffx = abs(xCentroids - columns/2);
[minXDiff, index] = min(diffx);
centerBlob2 = ismember(labeledImage, index);

I haven't tested that. It's just off the top of my head. If you run into trouble, write back.

Related Solutions

MATLAB: Imbinarize vs im2bw

label2rgb(label) is equivalent to label2rgb(label, 'jet') which triggers 'jet' to be executed with no parameters. When jet is executed, it checks to see if there is a current figure, and if so then it finds the number of rows of the figure's colormap, and if there is no current figure then it finds the number of rows of the graphics root object DefaultFigureColormap . Historically for Windows the default figure colormap had 64 rows, but if I recall correctly that was changed to 256, I think it was R2019a. Therefore the meaning of the label2rgb line depends upon how many colors are in the current colormap. It would be better if the code were rewritten to be independent of the current colormap.

The colormap generated by jet() is a double() colormap. But if the label array happens to be uint8 or uint16 or uint32 datatype, then it converts the colormap to the same uint* datatype . Unfortunately at the moment we cannot tell whether your label array is double or uint8.

im2bw(), presented with RGB data that is double(), does an rgb2gray() on the data, and then compares the results to the threshold, with the threshold defaulting to 0.5.

imb2bw(), presented with RGB data that is uint8, does an rgb2gray() on the ata, and then compres the results to the threshold, with the threshold defaulting to half of the maximum of the integer class (so 128 for uint8)

Therefore, you can replace your

    bw_img = im2bw(rgb_img);

with

if isfloat(rgb_img)
    bw_img = rgb2gray(rgb_img) > 0.5;
else
    bw_img = rgb2gray(rgb_img) >= 128;
end

which you can then optimize if you know the datatype of your label array.

Now, what is being tested is the brightness of the color induced by the label according to the jet colormap. The jet colormap happens to be such that for 256 colors, the color of the first entry is brighter than 1/2, and the color of entries 76 to 203 is greater than 1/2, and the colors of entries 2 to 75 and 204 to 256 are less than 1/2. This would pick out the first cluster, and clusters 76 to 203, and nothing else.

Your code

    label(label == 2) = 1;
    label(label == 3) = 2;
    label(label == 4) = 2;

hints that your data range in your label matrix is possibly 0 to 4 before the assigmments, and possibly 0 to 2 after the assignments. Or maybe it is 1 or 2 after the assignments, it is hard to tell from the snippet. But if so, if nothing in the label matrix is greater than 2 after those lines, then why go through all the trouble of doing the label2rgb and im2bw? Why not just say

    bw_img = label == 1;

or in terms of the label image before those three assignment statements,

    bw_img = ismember(label, [1 2]);

MATLAB: Mutilspectral image to RGB

To simply put each image into one of the R, G or B-layers of an RGB-image just do this:

img = peaks(12); % just something to give 
img1 = img;      % us three different
img2 = img1*0.7; % images with different
img3 = img2*0.7; % intensities
C1 = 1; % Possible 
C2 = 1; % intensity
C3 = 1; % scaling-factors
rgbimg = cat(3,img1*C1,img2*C2,img3*C3);

Now matlab expects RGB-images in double format to be between 0 and 1 to display them with functions like imagesc, so you might want to rescale rgbimg:

rgbimg = (rgbimg - min(rgbimg(:)))/(max(rgbimg(:)) - min(rgbimg(:)));

HTH

Best Answer

Related Solutions

MATLAB: Imbinarize vs im2bw

MATLAB: Mutilspectral image to RGB

Related Question