[ome-users] Checksum functionality, feature request.

[Josh Moore]

Hi Jake,

On Mon, Jan 4, 2016 at 3:59 AM, Jake Carroll <jake.carroll at uq.edu.au> wrote:
> Hi OME crew. Happy new year.

The same to you!

> I’ve been brain storming with colleagues over the last few hours and we had
> an interesting idea from a ‘digital fingerprint’ perspective.
>
> We like the idea of a checksum (perhaps an MD5 or an SHA) for each image or
> image set. We believe it might be a functional and useful feature in terms
> of digital preservation or “fixity” semantics and might be useful for proof
> of ownership/attribution principles down the track.
>
> Considering it, an MD5 on every data-set would be computationally fairly
> time consuming, so it’d definitely be a feature one could either turn on or
> off depending upon their circumstances and the computational power available
> to them (could have a boolean inside the omero config xml for whether it is
> turned on or off?),

As you likely know, what OMERO currently provides is a fileset-level
checksumming. At import time, a checksum can be chosen from the the
list of checksums that are supported by a particular server. (See [1])
All the files which make up a fileset *as Bio-Formats sees it* have
the same checksum-type calculated.


> ... but I’d love to see an extra information pane inside the
> “General” tab of the OMERO.web or Insight viewer detailing the checksum of
> an image or a set of images.

These can be viewed post import view the view button (See [2] from
[3]). But this is again, only for the files.

What you're suggesting has been discussed before on the list (e.g.
[4]) but hasn't been implemented. OMERO4 (i.e. pre-OMERO.fs) provided
the calculateMessageDigest [5] method which did something similar to
what you're suggesting though: loops through all planes in a given
order and hash them together. Again, that used sha1.


> We like the idea because it is effectively a mechanism of generating a
> digital watermark or fingerprint that does not have an impact on an image
> from a computational manipulation or transform perspective. It is also
> another arrow in the quiver of meta-data capability for OMERO.
>
> Something for consideration that we’d love to see in the future.

Understood and certainly no major objections. My primary question
since with OMERO.fs the entire fileset is needed to read out all the
planes in order to generate the checksum, what are the primary
benefits to the plane-based checksum from your perspective?


> Thank you for your consideration.
> -jc

And thanks for your continued feedback & support!
~Josh.


[1] https://www.openmicroscopy.org/site/support/omero5.2/sysadmins/config.html#omero-checksum-supported
[2] http://help.openmicroscopy.org/images/gettingStarted5Close.jpg
[3] http://help.openmicroscopy.org/getting-started-5.html#adding-data
[4] http://lists.openmicroscopy.org.uk/pipermail/ome-devel/2014-February/002680.html
[5] https://github.com/openmicroscopy/openmicroscopy/blob/v5.2.0/components/server/src/ome/services/RawPixelsBean.java#L330