Image Processing Question

[Sebastian Rhode-2]

Hi,

here is what I do for a little plugin in Java:

pic1 - pic2 = result1 --> measure --> min1 = 0

(pic1 +100) - pic2  = result2 --> measure --> min2 = 53

Built-in Plugins give the same results. Is ImageJ setting negative Pixel
values to Zero, since I expected min2=100?

And of course if you do image statistics stdDev (result1) and stdDev
(result2), the results will be different. In my opinion the addition of an
offset is required for the correct calculation of stdDev, but maybe I am
wrong.
Any suggestions on this issue?

Thanks for your help!

Sebi

[Colin Rickman]

Hi,

Make sure you are using the correct bit depth for your images in these
calculations. I would recommend using 32bit to allow negative numbers
and prevent the maths failing due to the values hitting the top of the
available range (ie 180 + 100 = 255 in 8bit data).

Hope that helps


Dr Colin Rickman
Centre for Integrative Physiology
School of Biomedical Sciences
University of Edinburgh
Hugh Robson Building
George Square
Edinburgh
EH8 9XD

Tel: [phone]
Fax: [phone]

Sebastian Rhode wrote:

> Hi,
>
> here is what I do for a little plugin in Java:
>
> pic1 - pic2 = result1 --> measure --> min1 = 0
>
> (pic1 +100) - pic2  = result2 --> measure --> min2 = 53
>
> Built-in Plugins give the same results. Is ImageJ setting negative Pixel
> values to Zero, since I expected min2=100?
>
> And of course if you do image statistics stdDev (result1) and stdDev
> (result2), the results will be different. In my opinion the addition of an
> offset is required for the correct calculation of stdDev, but maybe I am
> wrong.
> Any suggestions on this issue?
>
> Thanks for your help!
>
> Sebi
>

--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

[Pang, Zhengyu (GE Global Research)]

Just answer the statistics part of your question. Adding/substracting a
constant number to your orignal data will not change the standard
deviation.




Hi,

here is what I do for a little plugin in Java:

pic1 - pic2 = result1 --> measure --> min1 = 0

(pic1 +100) - pic2  = result2 --> measure --> min2 = 53

Built-in Plugins give the same results. Is ImageJ setting negative Pixel
values to Zero, since I expected min2=100?

And of course if you do image statistics stdDev (result1) and stdDev
(result2), the results will be different. In my opinion the addition of
an offset is required for the correct calculation of stdDev, but maybe I
am wrong.
Any suggestions on this issue?
Thanks for your help!

Sebi

[Joachim Wesner]

Hi again,

JAIN, mathematically true, but 8-Bit and 16-Bit ImageJ image format
(if not scaled) is considered unsigned, so any negative calculation result
will be clamped to zero, you need to at least convert the image to 32-Bit
- or add an appropriate offset - to retain image statistics.

JW

ImageJ Interest Group <[hidden email]> schrieb am 07.05.2008 17:48:48:

> Just answer the statistics part of your question. Adding/substracting a
> constant number to your orignal data will not change the standard
> deviation.
>
>
> Hi,
>
> here is what I do for a little plugin in Java:
>
> pic1 - pic2 = result1 --> measure --> min1 = 0
>
> (pic1 +100) - pic2  = result2 --> measure --> min2 = 53
>
> Built-in Plugins give the same results. Is ImageJ setting negative Pixel
> values to Zero, since I expected min2=100?
>
> And of course if you do image statistics stdDev (result1) and stdDev
> (result2), the results will be different. In my opinion the addition of
> an offset is required for the correct calculation of stdDev, but maybe I
> am wrong.
> Any suggestions on this issue?
> Thanks for your help!
>
> Sebi

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________