Convolution Mask

To start, I have 3 matrices:

m1= [3 x 3]
m2= [5 x 5]
m2= [7 x7]

I have three loops that are used depending on the size of the matrix. I want to use an "if" command to decifer the size of matrix so the appropriate loop is used.

Here is an example of the 3 x 3 case.

function [matrix]= three_by_three(x,y,I)
if m1 = 
matrix= [I(x-1, y-1), I(x-1, y), I(x-1, y+1); I(x, y-1), I(x,y), I(x, y+1); I(x+1, y-1), I(x+1, y), I(x+1, y+1)];
end

Answer 1

[David Young]

Something like this?

sz = size(matrix);
if isequal(sz, [3 3])
  % first for loop, function call or whatever
elseif isequal(sz, [5 5])
  % second 
elseif isequal(sz, [7 7])
  % third 
else
  error('myfunction:badsize', 'Unexpected matrix size');
end

You could also use a switch statement.

Answer 2

[Image Analyst]

No need to have multiple functions to do this. A single line of code will extract any size submatrix you want. It all comes down to a basic MATLAB instruction like

subMatrix = fullSizeMatrix(row1:row2, col1:col2);

Here's a full demo:

% Generate sample data.
matrix = floor(6*rand(15))+1
[rows columns] = size(matrix)

% Define center of submatrix to extract.
row = 4;
column = 6

% First example: Extract a 3x3 chunk.
sizeToExtract = 3;
halfWidth = floor(sizeToExtract / 2)
subMatrix = matrix(row-halfWidth:row+halfWidth, column-halfWidth:column+halfWidth)

%Second example: Extract a 5x5 chunk.
sizeToExtract = 5;
halfWidth = floor(sizeToExtract / 2)
subMatrix = matrix(row-halfWidth:row+halfWidth, column-halfWidth:column+halfWidth)

By the way, you do know that there is a function called conv2() that does convolution so you don't need to reinvent it, don't you?

[Joseph]

Yes, thanks for the response. However, we had to write our own for a class.