Re: STANDARD DEVIATON

[Ben Marriage <ben(at)met.ed.ac.uk>]

Struan Gray wrote:
 root.
> 
> 
>     function imageSD, image
> 
>       localmean = smooth(float(image), 3, /edge_truncate)
>       localsd = (float(image)-temporary(localmean))^^2
>       localsd = smooth(temporary(localsd), 3, /edge_truncate)
>       localsd = sqrt(temporary(localsd))
> 
>       return, localsd
> 
>     end
> 
> 
> Struan

I believe that this gives different results to the inbuilt IDL routines.
The problem is a wee bit more complicated than it looks. 

My routine:

;=============================================================
function ben_imageSD, image

siz = size(image,/dim)

standdev = fltarr(siz[0],siz[1])

for i=1,siz[0]-2 do begin
    for j=1,siz[1]-2 do begin
        standdev[i,j] = sqrt(total((image[i-1:i+1,j-1:j+1]-$
                                 
total(image[i-1:i+1,j-1:j+1])/9.)^2)/8.)
    endfor            ;tricky bit ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
endfor                ;to do without loops
return,standdev                
end
;=============================================================

IDL's version:

;=============================================================
function IDL_imageSD, image

siz = size(image,/dim)

standdev = fltarr(siz[0],siz[1])

for i=1,siz[0]-2 do begin
    for j=1,siz[1]-2 do begin
        standdev[i,j]=stddev(image[i-1:i+1,j-1:j+1])
    endfor
endfor
return,standdev
end
;=============================================================

They do similar things but with loops. Can't figure out how to do this
without loops!

Thanks,
Ben