Hatch pattern inside connected components in a binary image

Hi everybody!

I have attached a binary matrix (.mat file attached), which looks something like this:

I would like to put hatch pattern inside the two curves, something like this:

I looked at this answer, but couldn't use it to my advantage.

I also have the means of demarcating the different regions, something like this:

As you can say, the regions inside, outside, and on the curve have different cell values. I can demarcate the region, but I am not able to fill the "inside" region with a hatched pattern. The code to obtain the third image from the first image is attached below (it may be an inefficient code, but efficiency isn't my primary concern right now).

 %% Separate binary matrix into 2 parts
    BinMat = mydata;
    BinMat_left = BinMat(:,1:end/2);
    BinMat_right = BinMat(:,(end/2)+1:end);
    %% Processing for left part
    BinImage_right = ~logical(BinMat_left);
    % Get a marker image to reconstruct just the connected region and not all the other disconnected regions.
    binaryMarkerImage_left = false(size(BinMat_left));
    % Specify the starting point - a marker.
    row = 1;
    column = 3;
    binaryMarkerImage_left(row, column) = true;
    reconImage_left = imreconstruct(binaryMarkerImage_left, BinImage_right, 4);
    filled_win_left = BinMat_left + 2 * reconImage_left;
    filled_win_left = imbinarize(filled_win_left);
    filled_win_left = imcomplement(filled_win_left);
    l_left = bwlabel(filled_win_left,4);
    [r_left, c_left] = find(l_left==2);
    indices_left = [r_left,c_left];
    for len_left = 1:length(r_left)
        l_left(indices_left(len_left,1),indices_left(len_left,2)) = 0;
    end
    dil_ima_2_left = l_left;
    edge_corrected_left1 = edge(l_left,'canny');
    edge_corrected_left = bwareaopen(edge_corrected_left1,2);
    % Demarcating regions on either side of flame edge
    I_left = double(edge_corrected_left); % BW2 is the logical matrix with flame edge data
    [m_left,n_left] = size(I_left);
    [xx_left,yy_left] = find(I_left); % find coordinates of flame edge which will work as nodes
    % dil_ima_2 = flip(dil_ima_2,1); % dil_ima_2 is the binarized image obtained PLIF image
    for i_left = 1:m_left
        for j_left = 1:n_left
            if (dil_ima_2_left(i_left,j_left)==0)
                I_left(i_left,j_left)=-3;
            end
            if (dil_ima_2_left(i_left,j_left)==1)
                I_left(i_left,j_left)=-2;
            end
        end
    end
    II_l = double(edge_corrected_left);
    % This loop assigns -3 to region outside the flame edge, 1 to flame edge & -2 to region inside the flame edge in I_l
    for i_left = 1:m_left
        for j_left = 1:n_left
            if(II_l(i_left,j_left)==1)
                I_left(i_left,j_left) = II_l(i_left,j_left);
            end
        end
    end
    %%  Processing for right part
    BinImage_right = ~logical(BinMat_right);
    % Get a marker image to reconstruct just the connected region and not all the other disconnected regions.
    binaryMarkerImage_right = false(size(BinMat_right));
    % Specify the starting point - a marker.
    row = 1;
    column = 3;
    binaryMarkerImage_right(row, column) = true;
    reconImage_right = imreconstruct(binaryMarkerImage_right, BinImage_right, 4);
    filled_win_right = BinMat_right + 2 * reconImage_right;
    filled_win_right = imbinarize(filled_win_right);
    filled_win_right = imcomplement(filled_win_right);
    l_right = bwlabel(filled_win_right,4);
    [r_right, c_right] = find(l_right==2);
    indices_right = [r_right,c_right];
    for len_right = 1:length(r_right)
        l_right(indices_right(len_right,1),indices_right(len_right,2)) = 0;
    end
    dil_ima_2_right = l_right;
    edge_corrected_right1 = edge(l_right,'canny');
    edge_corrected_right = bwareaopen(edge_corrected_right1,2); % Remove isolated non-connected pixels from all images
    % Demarcating regiins inside and outside the flame edge
    I_right = double(edge_corrected_right);
    [m_right,n_right] = size(I_right);
    [xx_right,yy_right] = find(I_right); % find coordinates of flame edge which will work as nodes
    % dil_ima_2 = flip(dil_ima_2,1); % dil_ima_2 is the binarized image obtained PLIF image
    for i_right = 1:m_right
        for j_right = 1:n_right
            if (dil_ima_2_right(i_right,j_right)==0)
                I_right(i_right,j_right)=-3;
            end
            if (dil_ima_2_right(i_right,j_right)==1)
                I_right(i_right,j_right)=-2;
            end
        end
    end
    II_right = double(edge_corrected_right);
    % This loop assigns -3 to products, 1 to flame edge & -2 to reactants in  I_r  
    for i_right = 1:m_right
        for j_right = 1:n_right
            if(II_right(i_right,j_right)==1)
                I_right(i_right,j_right) = II_right(i_right,j_right);
            end
        end
    end
    %% Full I image
    I = [I_left I_right];
    imagesc(I)

Can someone please help me with this? Also, if it isn't too much trouble, I would like to overlay this hatched image over some other data, something like this:

In this image, I overlay the binary image on a different image. To accomplish this best, I think the best approach would be to fill the region (which is to be hatched) with 1s in a diagonal fashion (in the binary image), such that the 1s themselves will give a hatched pattern. But I haven't been able to do this.

I request someone to please help. Any way forward also would be extremely useful. Thank you.

[Walter Roberson]

what is the reason you cannot use https://www.mathworks.com/matlabcentral/answers/417110-how-to-fill-2-d-closed-loops-with-lines-boundary-of-this-closed-loops-are-extracted-as-output-from#answer_335031

??

You can use one of the boundary tracing functions to get the point lists.

[Matt J]

Francesco Pignatelli's comment moved here:

Hello Ankit,

I am facing a similar problem, did you solve it in the end?

I am also struggling on generating a file similar to your 'BinMat.mat' since I did not understand how to extract an edge from my PLIF data. I think we have a similar situation, I guess your data come from a swirling flame.

Could you please share with me your knowledge about how to get the edge binarizing a a matrix?

Best

Francesco

[Ankit Sahay]

Hi Francesco,

I apologize for the late reply. I was not able to solve it, although primarily the reason might have been that I had to postpone the project, and I have not been able to pick it up until now. It has been quite a time since I have attempted to solve this, maybe some fresh ideas will help me solve it. I will let you know about the results.

You are right, my data comes from a swirl combustor. For edge detection, I initially use 2D-median filtering (medfilt2), followed by a combination of operations using adaptthresh, imfill, imdilate, and bmorph to detect the edges. If you give me some time, I will try to find a working algorithm to include here along with a sample data.

Again, I apologize for not following up on this problem, I should've. I will pick this up again as soon as I have some time (PhD is tougher than I expected!!). Please let me know if you have any more questions.

[Matt J]

Hi Francesco, I apologize for the late reply. I was not able to solve it,

@Ankit Sahay Puzzling. I gave you a solution in my answer below which seems to work as requested. You made no comment on it at all.

[Francesco Pignatelli]

Hello @Ankit Sahay, no worries at all for the delay, it is totally ok :) Thank you very much! Looking forward to see your algorithm. Also I am still working on it but I realized I have an issue on binarizing so I am still one step behind this issue. Once that is fixed I will keep on finding the solution related to the issue of this post and I will also try the @Matt J solution and I will come back here with a possibile solution (if I can find it). Again, thank you very much :)

Answer 1

[Matt J]

Perhaps as follows,

load Regions

[m,n]=size(BW);

N=7; %controls hatch density

hatch = ( mod((1:m).'+ (1:n),N) == (N-1) );

border=BW & ~imerode(BW,ones(3));

BW=BW & hatch | border;

imshow(BW)