How to find guassian gradient of a matrix with many Nan values

I have a matrix with 40 x 60 pixels with many Nan values ,When I applied guassian gradient ,result is Nan matrix.

GHow to ignore Nan values for applying guassian gradient.

Answer 1

[Image Analyst]

I'd use a modified median filter. So

% Set all nan values to inf, take the median filter of the array.
nanIndexes = isnan(m);
m(nanIndexes) = inf;
mfm = medfilt2(m, [3,3]); % Make window big enough to cover any nan regions.
% Then replace nan's in the original matrix with the median filtered values:
m(nanIndexes) = mfm(nanIndexes)

Then do your gradient operation, like with imgradient() or whatever. Try that and see if it works. If it doesn't, attach your data and code.

[Suchitra Ojha]

Analyst,

Thank you for the response .

I have already used adaptive median filter on the image.My values are in matrix form but when I am using 2nd line of your code ,it is a vector form with zero values.Even after this code imgradient gives Nan values.

[Image Analyst]

I don't know how. m and nanIndexes are both 2-D matrices, not vectors of zero, as long as there are nan's in the m matrix. Here's a small example

m = magic(7);
m(4,4) = nan % Set the middle pixel to nan.
% Set all nan values to inf, take the median filter of the array.
nanIndexes = isnan(m)
m(nanIndexes) = inf
mfm = medfilt2(m, [3,3]) % Make window big enough to cover any nan regions.
% Then replace nan's in the original matrix with the median filtered values:
m(nanIndexes) = mfm(nanIndexes)

First you'll see

m =

30 39 48 1 10 19 28

38 47 7 9 18 27 29

46 6 8 17 26 35 37

5 14 16 NaN 34 36 45

13 15 24 33 42 44 4

21 23 32 41 43 3 12

22 31 40 49 2 11 20

then after repair you'll see

m =

30 39 48 1 10 19 28

38 47 7 9 18 27 29

46 6 8 17 26 35 37

5 14 16 26 34 36 45

13 15 24 33 42 44 4

21 23 32 41 43 3 12

22 31 40 49 2 11 20

Save your original matrix in a .mat file

save('answers.mat', 'm');

then attach here with the paper clip icon.

[Suchitra Ojha]

In my image its like

m=

Nan 39 48 Nan 10 19 28

38 47 7 9 Nan Nan Nan

46 Nan 8 17 Nan 35 Nan

Nan Nan 16 26 34 36 Nan

Nan Nan Nan 33 42 Nan 4

21 23 Nan 41 Nan 3 Nan

22 31 Nan 49 2 Nan Nan