How to pass a 3d array double[][][] from java to a Matlab function

I have a matlab function inside a jar package which is supposed to receive from my java class a double[][][]. Due the difference in the way java and matlab understand array structures when the array arrives to the matlab side it is messed.

Is there a way to convert java 3d array so then each dimension can be read like: (:,:,1) (:,:,2) and (:,:,3) in the matlab function?

Answer 1

[James Tursa]

As long as it arrives as a 3D variable but the dimensions seem backwards, perhaps you only need to permute it. E.g.,

x = 3D array from your java
x = permute(x,[3 2 1]); % permute to MATLAB memory ordering

[Felipe Reis]

@James thanks for your answer. It was the right start to achieve what I wanted. In my case I had to work it out like this: DS is my java double[][][] and from matlab i would like to access each chunk of data like

DS(:,:,indexchunkofdata)

and have a NxM matrix as result so i did:

PDS = permute(DS,[3 2 1])
ChunkOfData = PDS(:,:,indexChunkOfData)
ChunkOfData = ChunkOfData'

So it's kind of the last dimension after the permutation still needed some transpose to invert what was column into lines and vice-versa. Also a side comment is that after the permute new zeros columns where added to the last dimension!? Still don't know the reason. Any insight?

Thanks again

[James Tursa]

Try permute(DS,[2 3 1]) to get rid of that extra transpose at the end.

The permute function does not add any elements to the input array. So if you are getting extra 0's they are there before the permute operation.