| Derivative along one dimension of a data cube? [message #55080] |
Sat, 28 July 2007 09:30  |
![Ryan[1] is currently offline](theme/default/images/xoffline.png.pagespeed.ic.XRkd1fkXye.png "Ryan[1] is currently offline")
Ryan[1]
Messages: 2
Registered: July 2007
|
Junior Member |
|
|
Hi folks,
I am working with some hyperspectral images and would like to take the
derivative of all the spectra in the cube. Is there a faster way to do
this than looping through the x,y dimensions and calling deriv on
every individual wavelength dimension?
Thanks!
Ryan
|
|
|
|
| Re: Derivative [message #60985 is a reply to message #55080] |
Tue, 01 July 2008 12:36  |
Vince Hradil
Messages: 574
Registered: December 1999
|
Senior Member |
|
|
On Jul 1, 2:11 pm, w...@bao.ac.cn wrote:
> On Jul 2, 3:05 am, Vince Hradil <hrad...@yahoo.com> wrote:
>
>
>
>> On Jul 1, 1:58 pm, w...@bao.ac.cn wrote:
>
>>> I have a 2-D array (a image) A. I want to know its partial
>>> derivatives along X direction and Y direction.
>>> dA(x,y)/dx=B
>>> dA(x,y)/dy=C
>>> B and C should also be 2-D array.
>>> I want to know if IDL has some function or operator can do this job
>>> directly.
>>> If it has not those I wanted, then I must write some program according
>>> to "Spline and Lagrange".
>>> I am not a foolish boy. But I am a lazy boy.
>>> Who can tell me if IDL can do this job for me?
>>> Thanks!
>
>> have you tried deriv()? Have you tried:
>> IDL> ?
>
> yes,I have used deriv() before.But it only works for one dimension.
> If I use it to differentiate a 2-D image along one direction, then I
> must use "for i= , do begin .....endfor". That is too slow.
> Could you give me more hints?
Ahh... sorry - I was in a bad mood.
This is what I've done (as part of a Canny routine, btw):
isize = size(image,/dimensions)
ncol = isize[0]
nrow = isize[1]
grad = fltarr(ncol,nrow,4)
filter = double([ [-1,-2,-3,-2,-1], [0,0,0,0,0], [1,2,3,2,1] ])
grad[*,*,0] = convol(image,transpose(filter),/center,/edge_truncate) ;
Horizontal
grad[*,*,1] = convol(image,filter,/center,/edge_truncate) ; Vertical
grad[*,*,2] = sqrt( grad[*,*,0]^2 + grad[*,*,1]^2 ) ; Magnitude
grad[*,*,3] = atan( grad[*,*,1], grad[*,*,0] ) ; Direction
I hope this helps you get started. I might have horiz/vert
reversed... it's been a while since I actually used this 8^)
|
|
|
|
| Re: Derivative [message #60986 is a reply to message #55080] |
Tue, 01 July 2008 12:11 |
wxf
Messages: 6
Registered: March 2007
|
Junior Member |
|
|
On Jul 2, 3:05 am, Vince Hradil <hrad...@yahoo.com> wrote:
> On Jul 1, 1:58 pm, w...@bao.ac.cn wrote:
>
>> I have a 2-D array (a image) A. I want to know its partial
>> derivatives along X direction and Y direction.
>> dA(x,y)/dx=B
>> dA(x,y)/dy=C
>> B and C should also be 2-D array.
>> I want to know if IDL has some function or operator can do this job
>> directly.
>> If it has not those I wanted, then I must write some program according
>> to "Spline and Lagrange".
>> I am not a foolish boy. But I am a lazy boy.
>> Who can tell me if IDL can do this job for me?
>> Thanks!
>
> have you tried deriv()? Have you tried:
> IDL> ?
yes,I have used deriv() before.But it only works for one dimension.
If I use it to differentiate a 2-D image along one direction, then I
must use "for i= , do begin .....endfor". That is too slow.
Could you give me more hints?
|
|
|
|
| Re: Derivative [message #60987 is a reply to message #55080] |
Tue, 01 July 2008 12:05 |
Vince Hradil
Messages: 574
Registered: December 1999
|
Senior Member |
|
|
On Jul 1, 1:58 pm, w...@bao.ac.cn wrote:
> I have a 2-D array (a image) A. I want to know its partial
> derivatives along X direction and Y direction.
> dA(x,y)/dx=B
> dA(x,y)/dy=C
> B and C should also be 2-D array.
> I want to know if IDL has some function or operator can do this job
> directly.
> If it has not those I wanted, then I must write some program according
> to "Spline and Lagrange".
> I am not a foolish boy. But I am a lazy boy.
> Who can tell me if IDL can do this job for me?
> Thanks!
have you tried deriv()? Have you tried:
IDL> ?
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
