| Padding arrays - vector subscripts not working [message #71586] |
Fri, 02 July 2010 11:45  |
![James[2] is currently offline](theme/default/images/xoffline.png.pagespeed.ic.XRkd1fkXye.png "James[2] is currently offline")
James[2]
Messages: 44
Registered: November 2009
|
Member |
|
|
Hello Everyone,
I'm writing a phase correlation routine for image registration. A
preliminary step is padding the images with zeroes. Here's a
simplified example of what I'd like to do:
X = bytarr(5,5,5)
Y = bindgen(2,2,3)
diff = (size(X, /dimensions) - size(Y, /dimensions))/2
X[diff] = Y
Unfortunately, this gives me an error message:
% Array subscript for X must have same size as source expression.
The strange part is, if I type in:
X[1,1,1] = Y
it works fine - but using the array DIFF (whose contents are [1,1,1])
it fails. What's happening? Is there any way I can do this in IDL?
By the way, I would like my final program to work on arrays with any
number of dimensions, so I'd rather avoid a kludge like X[diff[0],
diff[1], diff[2]] = Y.
|
|
|
|
| Re: Padding arrays - vector subscripts not working [message #71642 is a reply to message #71586] |
Wed, 07 July 2010 11:20  |
James[2]
Messages: 44
Registered: November 2009
|
Member |
|
|
The following case-statement method is pretty ugly, but it is good for
speed and memory use. I tested your program and it works, but it is
slower than the program below because of its L64INDGEN call.
type = size(array, /type)
out = make_array(dims, type=type, value=filler)
x = (dims - sz)/2
case numdims of
1: out[x[0]] = array
2: out[x[0], x[1]] = array
3: out[x[0], x[1], x[2]] = array
4: out[x[0], x[1], x[2], x[3]] = array
5: out[x[0], x[1], x[2], x[3], x[4]] = array
6: out[x[0], x[1], x[2], x[3], x[4], x[5]] = array
7: out[x[0], x[1], x[2], x[3], x[4], x[5], x[6]] = array
8: out[x[0], x[1], x[2], x[3], x[4], x[5], x[6], x[7]] = array
endcase
return, out
|
|
|
|
| Re: Padding arrays - vector subscripts not working [message #71647 is a reply to message #71586] |
Tue, 06 July 2010 13:48 |
penteado
Messages: 866
Registered: February 2018
|
Senior Member
Administrator |
|
|
I think this does it, but you should check. I am assuming that both
out and array have the same number of dimensions.
function pad, array, dims
;error checking, etc. removed for more compact newsgroup post
type = size(array, /type)
out = make_array(dims, type=type)
;calculate where to put the original array in the new
sz = size(array, /dimensions)
dimdiff=dims-sz
diff = dimdiff/2
out_strides = [1L, product(dims, /integer, /cumulative)]
out_ind = total(out_strides*diff, /integer) ;starting index where
out is to be set
array_ind=l64indgen(n_elements(array)) ;indexes of array to use in
out
out_ind+=array_ind ;make the index list a contiguous list starting
at out_ind
;if more than 1D, the section of out to be set may not be
contiguous, so make the jumps in the indexes at every dimension
array_strides = [1L, product(sz, /integer, /cumulative)]
for i=1,n_elements(dims)-1 do out_ind+=(array_ind/
array_strides[i])*dimdiff[i-1]*out_strides[i-1]
out[out_ind] = array[array_ind]
return, out
end
|
|
|
|
| Re: Padding arrays - vector subscripts not working [message #71650 is a reply to message #71586] |
Tue, 06 July 2010 12:15 |
penteado
Messages: 866
Registered: February 2018
|
Senior Member
Administrator |
|
|
On Jul 6, 3:38 pm, James <donje...@gmail.com> wrote:
> out[start] = array
Sorry, that was meant to be
out[start]=array[*]
However, this lead to me noticing that the part of out you want to set
is not contiguous. So start will have to be an array containing the
proper indexes of out that correspond to the locations to be changed,
instead of just the starting index. Which is a bit tricky for an
arbitrary number of dimensions.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
