| Re: coherence test implementation [message #15328] |
Thu, 13 May 1999 00:00 |
eddie haskell
Messages: 29
Registered: September 1998
|
Junior Member |
|
|
Mark Rehbein wrote (paraphrased):
>
> I'm doing what some people call a coherence test.
> I have implemented this the following way:
>
> for y=1, lines-1 do begin
> for x=1, pixels-1 do begin
> matrix=ch4(x-1:x+1, y-1:y+1)
> stats=moment(matrix, sdev=sdev)
> sddevimage(x,y)=sdev
> endfor
> endfor
>
> Can the code be more efficient if I use array and matrix operations?
Mark,
I took a shot at it, tried a couple ideas, including reordering the
original array into a [9,x] array (multiple elements where needed) and
doing a stdev calculation on all the rows simultaneously. In addition
to enlarging the original array by a factor of almost 9 (bad), it only
seemed to slow things down (also bad).
If you only need the standard deviation of each submatrix, you can save
a lot of time by only calculating that particular moment directly:
for y=1,lines-1 do for x=1,pixels-1 do begin
matrix=ch4[x-1:x+1,y-1:y+1]
n = n_elements(matrix)
sddevimage[x,y]=sqrt(total((matrix-(total(matrix)/n))^2)/(n- 1))
endfor
When I tested this it ran 10x faster than the case using moment, which,
whilst not what you asked for, is a significant speed increase.
Cheers,
eddie
----- ---- --- --- ---- --- -- --- --- -- -- - - - -
|\ A G Edward Haskell
|\ Center for Coastal Physical Oceanography
|\ Old Dominion University, Norfolk VA 23529
|\ Voice 757.683.4816 Fax 757.683.5550
|\ e-mail haskell*ccpo.odu.edu
----- ---- --- ---- --- --- --- --- -- -- -- - - - -
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
