Odd mask files after processing

View previous topic View next topic Go down

Odd mask files after processing

Post  jensr on Thu Nov 19, 2009 5:56 pm

Hello all

We have performed some scans of larval fish here in Aberdeen that seemed to work fine. An analyst have worked through images, performing manual line measurements of the larval fish in the images, and we've now run the batch processing over the same images in order to try and create conversion factors between the measurements automatically obtained and those directly measured.
However, the scan looks like there are no problems - no stripes or other problems during the scan, and a visual inspection of the image seems like it is ok (here's a resized pic of the scan):



However after we ran the conversion & processing, we noticed that the zooscan had only picked up about 50 out of 70 fish larvae in the image. When investigating the mask file, it seems something very odd has happened during the conversion or background subtraction:



The background scan looks clean:



So I'm wondering if this is a processing setting of if we have a(nother) major problem with the scanner?
avatar
jensr

Posts : 8
Join date : 2008-09-26
Age : 43
Location : Marine Scotland Science, Aberdeen, UK

View user profile http://zooimage.overchord.net

Back to top Go down

Pb with background

Post  picheral on Sun Nov 22, 2009 10:55 pm

Dear Jens,
I had a look at your images. They are apparently OK but I know that the eyes are much more adaptable than the machines !
I would like to get the image and associated files to check the process on my side and understand the reason which prevents a good background removal.
Regards
Marc P.

picheral

Posts : 60
Join date : 2008-05-02

View user profile http://www.zooscan.com

Back to top Go down

Re: Odd mask files after processing

Post  jensr on Thu Nov 26, 2009 12:22 pm

Just in case anyone has experienced similar problems. Marc is currently looking over some of our images, so a final verdict is pending. But in the meantime, I have reprocessed the sample-set using the rolling ball background method instead, and it has improved the process enormously.
avatar
jensr

Posts : 8
Join date : 2008-09-26
Age : 43
Location : Marine Scotland Science, Aberdeen, UK

View user profile http://zooimage.overchord.net

Back to top Go down

Background subtract on Zooscan from Aberdeen

Post  picheral on Mon Dec 07, 2009 5:35 am

Hi Jens,
The background subtract problem arises when the optical properties (light, calibration...) of the systems have either not been well adjusted in factory or modified by user.
Your Zooscan comes from the first series from Biotom Company. As you opened your machine and modified the internal calibration strip, the cover light probably needs an adjustment to provide an easy background subtraction. This must be done by expert but the Biotoms models are not supported by Hydroptic (to be checked). Hydroptic models are much better constructed now and the company cares about the delivered systems in case of problems.
As you wrote, the Rolling ball methods can be a solution but the settings must be adjusted to avoid having square artefacts on big targets. In the case of the small fish larvae it works fine with a default setting of 100 at 2400 dpi.
You may also try to work with the large transparent frame. The background may be better removed with this frame size.
Regards
Marc P.

picheral

Posts : 60
Join date : 2008-05-02

View user profile http://www.zooscan.com

Back to top Go down

Re: Odd mask files after processing

Post  Sponsored content


Sponsored content


Back to top Go down

View previous topic View next topic Back to top

- Similar topics

 
Permissions in this forum:
You cannot reply to topics in this forum