How to convert a figure in matrix of pixel?

I have a picture and I want to convert this into a numeric matrix. I want to create a matrix of 1's and 0's where the 0's represent the orange pixels and the 1's represent the transparent pixels.

Can anybody help me?

[Voss]

Is this picture stored in an image file (e.g., a .bmp, .jpg, .png, etc.)? If so, can you upload a copy?

[Rik]

What is your source object? Is it an actual image? Does that include the axes? Should the area be cropped?

[SAC CSA]

Hi @Rik and @Benjamin, thank you for the comment. The image has an extension .png. The image represents a bifurcation of tubulation. Then, is a real image. The idea is obtain a matrix of 128x128.

[SAC CSA]

@Rik and @Benjamin. The Mathematica code for this procedure is relatively simple, but I would like to get a version of the code in matlab.

image = Import["https://i.stack.imgur.com/CQkvZ.png"]
data = ImageData[ChanVeseBinarize[image]] 

Answer 1

[Walter Roberson]

filename = 'https://www.mathworks.com/matlabcentral/answers/uploaded_files/834235/image.png';
img = imread(filename);
whos

  Name            Size                Bytes  Class    Attributes

  filename        1x79                  158  char               
  img           373x363x3            406197  uint8              

bw = imbinarize( rgb2gray(img) );
whos bw

  Name        Size              Bytes  Class      Attributes

  bw        373x363            135399  logical              

imshow(bw)

Notice that the result is logical datatype, with 1 for the white pixels and 0 for the black pixels. Depending exactly what you are doing, you might need to convert that to double precision -- for example if you are wanting to use it as Alpha data, you will need to convert to double precision.

[SAC CSA]

Hi @Walter Roberson, thanks for the kind and priceless work. Instead to generate a matrix bw that has a dimension of 373x363, could be generated a 128x128 matrix? In other words, how to control the output matrix dimension?

[Walter Roberson]

filename = 'https://www.mathworks.com/matlabcentral/answers/uploaded_files/834235/image.png';
img = imread(filename);
bw = imresize(imbinarize( rgb2gray(img) ), [128 128]);
imshow(bw)

However, I would suggest that you might want to crop the image before resizing.

bw = imbinarize( rgb2gray(img) );

vertcounts = sum(~bw,1);
%plot(vertcounts)
[~, vertidx] = findpeaks(vertcounts, 'npeaks', 4, 'minpeakheight', 300);
vert_inside = vertidx(2:3) + [1 -1];

horzcounts = sum(~bw,2);
%plot(horzcounts)
[~, horzidx] = findpeaks(horzcounts, 'npeaks', 4, 'minpeakheight', 300);
horz_inside = horzidx(2:3) + [1; -1];

bwcrop = bw(horz_inside(1):horz_inside(2), vert_inside(1):vert_inside(2));
cb = ~imclearborder(~bwcrop);
result = imresize(cb, [128 128]);
imshow(result)

Figuring out where the inside was, was more work than I expected... I do not use findpeaks() very often.

[SAC CSA]

Hi @Walter Roberson, thank you very much for working on the code above. It worked perfectly for a 2D figure.

Now, I tried to apply an similar protocol to a 3D figure, like the image 2 attached.

However, although the image depicts a 3D plot, there are only 2D (448x416) pixels. I'm trying to get a 3D matrix (128x128x128) of 1's and 0's where the 0's represent the branches' pixels and the transparent part is 1's inside of the box.

In this case, do you think it would be necessary to transform the image into a picture in 3D pixels to then get the matrix?

Could you help me to obtain the matrix, please?

Thank you so much in advance!

[Walter Roberson]

Depth cannot be calculated from a single 2d picture (unless there are cues such as color)