Mean of neighbor pixel

Hi guys,

I try to calculate the average intensity value of the neighbors of a pixel(either 8 or not). I am confused with the way matlab uses matrices(i-th,j-th position of a pixel is at j-th,i-th position of the matrix) and I have difficulties find the neighbors.

Thanks

Answer 1

[Matt J]

Just use convolution,

k=[1 1 1; 1 0 1; 1 1 1]/8;

averageIntensities = conv2(double(yourImage),k,'same');

[Efstathios Kontolatis]

The problem is I need the neighbors of each pixel later so I have to save them.

[Arghya Pathak]

"same" stands for what?

[Matt J]

You can read all about it here:

https://www.mathworks.com/help/matlab/ref/conv2.html#bvgtez6-shape

Answer 2

[Adam]

If you are finding all 8 neighbours then it doesn't make a difference which is x,y and rows, columns, but your confusion and that comes from the fact that a matrix is indexed as (row,column), whereas an image is (x,y) more usually where x is horizontal and is therefore represented by columns of the image.

The neighbours of a pixel are simply

neighbours = yourMatrix( yLoc-1:yLoc+1, xLoc-1:xLoc+1 )

for a pixel at ( xLoc, yLoc ) in a matrix called yourMatrix.

This will also include the point itself, but that is easy to get rid of and you didn't state what you want to do with the neighbourhood so it may or may not be better to keep the central point for it to retain its 2d structure.

[Efstathios Kontolatis]

I want to find the average intensity of the neighbors without using the central point. Also, I don't think that this solution will give me the neighbors on the borders of the image. I used this code

for i=1:height
  for j=1:width
      neighbours(i,j).ar=img(j-1:j+1, i-1:i+1);
  end
end

and I take the following message 'Subscript indices must either be real positive integers or logicals'. Img is a uint16 image. As you can see I want to save the neighbors of each pixel.

[Adam]

Yes, you have to deal with the borders differently, but that is not difficult.

If you don't want the middle point then just throw it away and collapse the neighbours to a 1d array since you will just be taking the mean anyway - e.g.

neighbours(5) = [];