Question

  function imOut = medianFilter(imIn,windowWidth)

     if mod(windowWidth, 2 ) == 0
          disp('Window has even size');
     return
     end

     imageSize = size(imIn);
     imOut = imIn;

     windowBreadth = (windowWidth - 1)/2;
  for m=windowBreadth+1:imageSize(1) - windowBreadth
    for n=windowBreadth+1:imageSize(2) - windowBreadth
      t1 = imIn(m-windowBreadth:m+windowBreadth,n-windowBreadth:n+windowBreadth);
      t2 = reshape(t1,windowWidth*windowWidth,1);
      t3 = median(t2);
      imOut(m,n) = t3;
    end
  end

my explanation:

the function medianFilter takes an image(imIn) as an input, and a width for the window of the median filter

then I’m not sure why we need the if statement

after that we take the size of the input image and save it in a variable called imageSize

then we copy all the values of imIn to imOut

then I’m all lost

what is windowbreadth? and isn’t breadth and width the same thing?

thanks!

Answer 1

The if conditional is checking that the window-size is odd; if it’s even then there is no “central” pixel.

windowbreadth seems like a bad name. But its meaning should be clear from the diagram below:

|<--B-->|
+---+---+---+---+---+
|   |   |   |   |   |
+---+---+---+---+---+
|   |   |   |   |   |
+---+---+---+---+---+
|   |   | * |   |   |
+---+---+---+---+---+
|   |   |   |   |   |
+---+---+---+---+---+
|   |   |   |   |   |
+---+---+---+---+---+
|<--------S-------->|

where S is the window-size, B is the window-“breadth”, and * denotes the “central” pixel.

So t1 is all the samples in the current window. t2 is those samples rearranged as a 1D vector rather than a 2D array.