How to run an if statement in a cell?

I have a cell containting 125 by 125 elements where each element contains a local 36 by 36 array of data. I am new to this sort of work and I am currently carrying out some data analysis. I want to be able to determine if any of the local data (so data within the 36 by 36 array) are greater than 10% of the local average, but this needs to be done for the whole cell. I can carry this procedure out separately but this is not efficient. Am I correct that an 'if' statement would be able to get the job done? Any help on this would be greatly appreciated.

[Alex Mcaulley]

Probably cellfun will help you.

What do you want exactly as output? A logical 36x36 array for each cell determining which element is greater than the 10% of the mean of the 36x36 elements?

[Daniel Tanner]

Yes that would be the output I would want.

Answer 1

[Alex Mcaulley]

A possible soution:

sol = cellfun(@(in) in > 0.1*mean(in(:)),A,'uni',0); %where A is your cell array

[Daniel Tanner]

That is perfect, thank you very much!

Is there a way to then output the final cell as a 125 by 125 logical where if any of the data in the 36 by 36 elements is over 10% of the mean? Or is this not possible?

Many thanks for your help so far.

[Alex Mcaulley]

It can be done, but It will be allways true (at least 1 element must be greater than the 10% of their local mean).

sol = cellfun(@(in) any(any(in > 0.1*mean(in(:)))),A);

[Daniel Tanner]

Thank you.

[Daniel Tanner]

Hi Alex, is it possible to put two limits within the cellfun? As in 10% either side of the mean?

[Alex Mcaulley]

Yes, just putting an '&' operator between conditions:

sol = cellfun(@(in) any(any(in > 0.9*mean(in(:)) & in < 1.1*mean(in(:)))),A);

[Daniel Tanner]

Thank you.

Finally, and I understand this may not be possible. But what I am essentially doing is treating the original data set as pixels. So every point of data is a pixel that has detected a photon on a detector screen. I initially had a 4500 by 4500 array which I managed to split into a 125 by 125 cell with each element containg 36 by 36 pixels thanks to your help. I am looking at finding the number of defected pixels on the whole device (a defected pixel is one with a signal level >10% away from the mean either side, so < 0.9*mean and >1.1*mean are defined as defected pixels). Using your help I have managed to find the number of defected elements, and in them I have found the locations of the defected pixels and added them up, but that section is hard coded and can be tedious.

I wondered if there was a way to count all the defected pixels using code instead of counting all the defected areas myself. For example, here is the code I used to find the defected pixels for one of the data sets:

L900 = cellfun(@(in) in > 0.9*mean(in(:)) & in < 1.1*mean(in(:)),PRNU900_Cell,'uni',0);
DeflectedPixelsTest = sum(length(find(L900{123,2}>0))+length(find(L900{78,13}>0))+length(find(L900{31,17}>0))+length(find(L900{4,24}>0))+length(find(L900{93,24}>0))+length(find(L900{22,25}>0))+length(find(L900{23,25}>0))+length(find(L900{11,39}>0))+length(find(L900{92,41}>0))+length(find(L900{120,43}>0))+length(find(L900{62,53}>0))+length(find(L900{76,65}>0))+length(find(L900{103,69}>0))+length(find(L900{11,79}>0))+length(find(L900{32,84}>0))+length(find(L900{106,114}>0))+length(find(L900{107,114}>0))+length(find(L900{55,117}>0))+length(find(L900{81,119}>0)));

So as you can see, there were 19 defected regions in the whole of my L900 cell and I located them and counted the number of defected pixels within each of them.

Do you know of a more simpler way to do this? Sorry if this was confusing, let me know if anything needs clearing up. Thanks!

[Stephen23]

"Do you know of a more simpler way to do this?"

Take a look at blockproc, histcount, histcounts2, etc..