How to replace cell and neighbors that are zero in a matrix with nan

Hello,

I am attempting to code a basic version of minesweeper on MATLAB and I've got it working good so far. The next step for me is for when the player selects a point on the game board that has no bomb and no bombs around it, it should display all of those empty squares around it. My thoughts are, to make sense to the player, it should replace the square you selected and the neighboring sqaures (including those that dont directly touch it - up until it does neighbor a bomb) into NaN. For example:

     0     0     0     0     0     0    
     0     0     0     0     0     0    
     0     0     0     0     0     0    % empty gameboard is shown to player
     0     0     0     0     0     0    
     0     0     0     0     0     0    
     0     0     0     0     0     0    
     0     0     0     0     0     0     
     
     
     % player selects row 2 column 2 and there are no bombs around it 
     NaN  Nan  Nan   1     0     0    
     NaN  Nan  Nan   1     0     0    
     NaN  Nan  Nan  Nan    1     0    % empty spaces up until the boxes that neighor bombs are displayed
     Nan  NaN  Nan   2     0     0    
     1    Nan  1     3     0     0    
     0    4    0     0     0     0    
     0    0    0     0     0     0

I've got the code working to display the boxes with neighboring bombs (i.e. the boxes with numbers in them) but I cant figure out how to replace the empty squares until they reach the boxes neighboring bombs into NaNs.

In case it helps, here's the code that figures out the moves, how you win, and how you lose. The mines are randomly placed.

while X==0% %while the player hasnt found a mine

coords = input('type the coordinates in brackets: ');
rows = coords(1);
columns = coords(2);
    
    if answer_matrix(rows,columns)==0          % answer matrix is matrix that contains bombs (0 means no bomb 1 means bomb)
        player_matrix(rows,columns)=neighbours(rows,columns); %player matrix is matrix show to player 
        disp(player_matrix)                                   %neighbors is matrix that contains every square with a number (neighboring bombs)
        win_matrix(rows,columns)=0;       
        moves = moves + 1
        
        if win_matrix==answer_matrix
            X=1;
            disp('you win!!') 
        end
        
         elseif answer_matrix(rows,columns)==1
        disp(answer_matrix);
        disp('oops that is a mine')
        break
  
    end
end

Answer 1

[Matt J]

Perhaps as follows:

A=[     0     0     0     1     0     0
        0     0     0     1     0     0
        0     0     0     0     1     0
        0     0     0     2     0     0
        1     0     1     3     0     0
        0     4     0     0     0     0
        0     0     0     0     0     0 ];
    
loc=[2,2];
    

    fillregion=imfill( A>0 ,loc,4) & A==0;
    
    A(fillregion)=nan

A = 7×6

NaN NaN NaN 1 0 0 NaN NaN NaN 1 0 0 NaN NaN NaN NaN 1 0 NaN NaN NaN 2 0 0 1 NaN 1 3 0 0 0 4 0 0 0 0 0 0 0 0 0 0

Answer 2

[DGM]

If you have IPT, this becomes a simple segmentation problem:

A = [0 0 0 1 0 0;
    0 0 0 0 1 0;
    0 0 0 2 0 0;
    1 0 1 3 0 0;
    0 4 0 0 0 0;
    0 0 0 0 0 0];

% get 4-connected label array
% this lists all contiguous blobs of true values in the input
L = bwlabel(A==0,4)

L = 6×6

1 1 1 0 2 2 1 1 1 1 0 2 1 1 1 0 2 2 0 1 0 0 2 2 2 0 2 2 2 2 2 2 2 2 2 2

% for a sample point, fill the given blob
pickpt = [2 2]; % row, col
A(L == L(pickpt(1),pickpt(2))) = NaN

A = 6×6

NaN NaN NaN 1 0 0 NaN NaN NaN NaN 1 0 NaN NaN NaN 2 0 0 1 NaN 1 3 0 0 0 4 0 0 0 0 0 0 0 0 0 0

[Matt J]

I wonder if this solution correctly handles the case when you take the 4 away...? I feel like there should be the same pattern of NaNs when the 4 is present since the NaN front stops when all the NaNs are 4-connected to a non-zero.

A=[     0     0     0     1     0     0
        0     0     0     1     0     0
        0     0     0     0     1     0
        0     0     0     2     0     0
        1     0     1     3     0     0
        0     0     0     0     0     0
        0     0     0     0     0     0 ];

L = bwlabel(A==0,4);
pickpt = [2 2]; % row, col
A(L == L(pickpt(1),pickpt(2))) = NaN

A = 7×6

NaN NaN NaN 1 NaN NaN NaN NaN NaN 1 NaN NaN NaN NaN NaN NaN 1 NaN NaN NaN NaN 2 NaN NaN 1 NaN 1 3 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN

[DGM]

It fills a 4-connected region of zeros, which is what I assumed the behavior would be based on the example. I don't remember how windows minesweeper behaved, but Gnomines appears to fill as I describe, but using 8-connectivity (i.e. even greedier yet).

Granted, the example is a bit misrepresentative, as no mine placement could create such a marker pattern, but if we're filling empty regions based on marker pixels, then the above is what I consider to be a reasonable extrapolation from familiarity toward the specific problem description at hand.

EDIT: Consider that this is the result of 1 click:

So then I would assume the classic behavior is as follows:

• if you click a bomb, you reveal a bomb; game ends
• if you click a marker pixel, you reveal the marker
• if you click a zero pixel, you reveal the 8-connected region of zeros and its boundary of marker pixels

But modifying the established behavior to use 4-connectivity instead seems like a simple enough alteration.

[Alex Jackson]

Thanks for the answer.

Do you think it is possible without using IPT? I'm trying to make it as simple as possible for the player, which includes downloading extra add-ons for the player to use to make the game functional.

[DGM]

Oof. I mean... anything's possible without IPT, but you'd have to implement similar functionality in your own code at the cost of increasing the complexity of your programming task.

That said, both of our examples use tools from IPT (imfill, bwlabel). A lot of simplified workarounds to avoid those two functions might use morphological tools which are also part of IPT.

If it's acceptable to require a different dependency instead of IPT, then MIMT has bwlabelFB(), which should serve as a direct replacement for bwlabel(). This is a wrapper for gp_bwlabel() with a minor tweak to match IPT bwlabel() behavior.

The underlying function gp_bwlabel() is from Alec Jacobson's geometry processing toolbox; both MIMT and GPT are available freely on the FEX:

https://www.mathworks.com/matlabcentral/fileexchange/53786-image-manipulation-toolbox

https://www.mathworks.com/matlabcentral/fileexchange/49692-gptoolbox

I've attached the three files necessary from both MIMT and GPT. With this, the same example can be done without IPT.

A = [0 0 0 1 0 0;
    0 0 0 0 1 0;
    0 0 0 2 0 0;
    1 0 1 3 0 0;
    0 4 0 0 0 0;
    0 0 0 0 0 0];

% get 4-connected label array
% this lists all contiguous blobs of true values in the input
L = bwlabelFB(A==0,4);

% for a sample point, fill the given blob
pickpt = [2 2]; % row, col
A(L == L(pickpt(1),pickpt(2))) = NaN

A = 6×6

NaN NaN NaN 1 0 0 NaN NaN NaN NaN 1 0 NaN NaN NaN 2 0 0 1 NaN 1 3 0 0 0 4 0 0 0 0 0 0 0 0 0 0

So you don't need to include the whole of MIMT or GPT, just those three files.

If you decide that you don't want any foreign support files, feel free to consider these three files as a reference implementation.