2D interpolation - how to deal with "missing" data

Hello!

I have a matrix with missing numbers. There are some rows and columns missing; these I can find by interpolating with interp2. However, there are also some gaps, i.e. "missing" data, in my matrix that can not be retrieved by the same method. These empty entries are automatically set to 0 (zero) when I import my data to Matlab, and are subsequently treated as such instead of treated as gaps which need to be "filled" when I try to use interp2.

I could do a 1D interpolation at each row, column, and/or diagonal where there are gaps, but I am afraid that this would give me less good results (as I only take two out of eight neighbours into account) and that it will take me a lot of time.

Is there any easy way to perform an interpolation to fill the gaps?

Any help would be much appreciated!

Thank you

/Tomas

[nancy gupta]

how you processed interpolation diagonally? Actually I am stuck with it.

Thank You

Answer 1

[Image Analyst]

How about using triscatteredinterp()?

Answer 2

[Walter Roberson]

Perhaps John's File Exchange contribution inpaint_nan would work for you.

[Tomas]

This did the trick - thank you very much!