Calculate mean value on a skeleton of an image

Hi everyone! I'm working on a skeleton of an image and I found some errors on it. I marked them with ones, so I can identify them easily. I had introduced data in the positions of the original skeleton, that's why most of the values are different to 1 and 0. I'd like to change the value of the ones by the mean value of the 3x3 neighbours of that position, that are different to 0 or 1. For example:

A = [0 0 0 265 0 ; 0 1 260 0 0  ; 0 250 0 0 0  ; 251 250 0 0 0 ; 251 0 0 0 0];

I'd like to change the 1 by 255, the mean (average) value of 260 and 250 (neighbours of 1 and with a different value of 1 and 0).

Answer 1

[Image Analyst]

What is A? The modified skeleton or the original image? A skeleton is a binary image of type logical. Ignoring your A, what I'd do is this:

% Blur image
blurredImage = conv2(double(grayImage), ones(3)/9, 'same');
% Replace image "B" with values of the blurred image at the location of the skeleton 1 values:
B(skelImage) = blurredImage(skelImage);

That's the best I can do because I can't follow what you're saying. You have a gray scale image, a logical skeleton image, and you even have a third image that I guess is the skeleton image converted from logical to gray scale and some "data" from the gray scale image has replaced some pixels (or something like that). So, with these 3 images, you only give the one, very-badly-named "A" image, and don't say exactly what it is. Please give all 3 matrices plus a fourth one that is your desired output image.

[Alyssa Webb]

Okay, so originally A is a skeleton of a path. After certain process I've changed the ones of A (paths) by the height of the ground in that point (coordinate Z). I found that not all of the points of the original skeleton have the coordinate Z, so that's why I've kept the value of 1 in some of them. It's not a gray scale image, sorry about that.

[Image Analyst]

OK, let's say the ground height image is called heightImage, rather than something bad like "B". So you'd do :

% Blur original height image
blurredImage = conv2(double(heightImage), ones(3)/9, 'same');
% Find out where the 1's in "A" live:
onesLocations = A == 1;
% Replace image "A" with values of the blurred image at the location of A's 1 values:
A(onesLocations) = blurredImage(onesLocations);

[Alyssa Webb]

Thank you so much!