Adjusting Spatial resolution

[Russell Crook]

I have a high resolution image and wish to simulate what the image would
look like if it were acquired using a low resolution camera with  1 meter
resolution at 500 meters for example.  Does anyone know how to do this?  I
suspect I need to use a convolution but am uncertain what function I should
use.  Any help would be appreciated.

 

Russ

 

 


--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Michael Schmid]

Hi Russ,

for an exact solution you have to know the point spread function (transfer function) of your optics, and convolve the image with it, or (this is equivalent) multiply the image with the modulation transfer function in Fourier space (FFT>Custom Filter)
All this assumes that the original image is infinitely sharp or at least much sharper than than the output.

http://en.wikipedia.org/wiki/Point_Spread_Function
http://en.wikipedia.org/wiki/Modulation_transfer_function

Not so exact solutions:
- Downsizing with averaging and bicubic interpolation
- Gaussian Blur

Note that  "1 meter resolution at 500 meters" is not very well defined.
Does it mean "pixel size corresponding to 1 m"?
Or "it can resolve a line pair with 1 m spacing"?
Or "it can split to objects with 1 meters distance"?
Or "modulation transfer function decreases to ...% at a spatial frequency of 1/meter"?

Michael
________________________________________________________________
On Jul 15, 2012, at 06:14, Russell A. Crook wrote:

> I have a high resolution image and wish to simulate what the image would
> look like if it were acquired using a low resolution camera with  1 meter
> resolution at 500 meters for example.  Does anyone know how to do this?  I
> suspect I need to use a convolution but am uncertain what function I should
> use.  Any help would be appreciated.
>
>
>
> Russ

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[David M Gauntt]

The simplest solution would be to resize the image (command
Image/Adjust/SizeŠ).  This would be appropriate if the original image were
taken by a camera with (for example) a 2048x2048 matrix size, and you want
to model a 512x512 matrix size, or you want an image taken at 100 meters
to simulate the central part of an image taken at 400 meters.

If the loss of resolution were instead due to an out-of-focus lens, then I
would recommend a gaussian blur.

--
David M. Gauntt, Ph.D.
Associate Professor,
Division of Medical Physics and Engineering
UAB Department of Radiology

mailto:[hidden email]
205-975-3777



On 7/15/12 11:00 PM, "Russell A. Crook" <[hidden email]> wrote:

>I have a high resolution image and wish to simulate what the image would
>look like if it were acquired using a low resolution camera with  1 meter
>resolution at 500 meters for example.  Does anyone know how to do this?  I
>suspect I need to use a convolution but am uncertain what function I
>should
>use.  Any help would be appreciated.
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html