How do I combine two index vectors to form a 2D index matrix?

I have two methods for identifying which elements in a matrix have certain properties (e.g.; exceed a threshold).

        ndx = Amat > THR;                  % matrix of threshold crossing locations

and:

        [rdx, cdx] = find(Amat>THR);       % vectors of threshold crossing locations

that I use for two different purposes [and actually apply to different matrices.] Results from the two methods are compared eventually. I'd like to take the two vectors (rdx, cdx) and convert them to a matrix so I can do an operation such as:

        Adx = Amat(ndx);          % matrix of threshold crossing amplitudes

which is done for the former case. I'm sure there's a simple way to take the two vectors and combine them to produce "pointer" matrix "ndx," but I haven't found it.

Help?

[Dean Ranmar]

Of course "ndx" is a true/false matrix of same dimension as matrix Amat.

Answer 1

[Jan]

Does not look elegant:

[rdx, cdx] = find(Amat > THR);
Index      = sub2ind(size(Amat), rdx, cdx);  % [EDITED, was ind2sub...]
ndx        = false(size(Amat));
ndx(Index) = true;

[Dean Ranmar]

Thanks! The key is the ind2sub function. Excellent.

[Dean Ranmar]

Actually, it should be sub2ind, not ind2sub. Thanks again. BTW, I should have neat rather than elegant and your solution is neat.

Answer 2

[Mischa Kim]

Would this do the trick?

 Adx = Amat(Amat(:)>THR(:))

[Dean Ranmar]

Yes, that does work in the sense it is equivalent to the 1st method - it combines my two steps into one line of code. Thanks. I was hoping for a simple way to start with the two index vectors and derive the Adx matrix you define above, from the vectors. It's really an academic exercise I guess, but I suspect one that will help me with MATLAB in the future. Right now, I combine the two vectors to produce a single pointing vector and reference the array with the single index vector:

        ndx = ndr*ncol + ndc;  % create vector to find thr crossings
        Adx = Amat(ndx);       % matrix of threshold crossings

Perhaps I need to create a mask array and fill it with zeros, then somehow use the two vectors to "true" the threshold crossing positions?

[Dean Ranmar]

ncol = number of columns in the array Amat.

[Dean Ranmar]

sorry! I switched the index names on you: ndr == rdx and ndc == rdc.

[Dean Ranmar]

ndc == cdx