Re: Collecting Results from Results Box

Posted by Brent Wood on Aug 29, 2005; 11:22pm
URL: http://imagej.273.s1.nabble.com/Collecting-Results-from-Results-Box-tp3704955p3704957.html

On Tue, 30 Aug 2005, Stella Sim wrote:

> Hi Brent,
>
> If you store the measurements  in the result table(default), the data will
> be appened to the result table everytime you run an image analysis anyway.
> What I have done is that I put all the images of interest in one folder,
> and run my image analysis for each image in the folder (automated proces),
> results for each image were then appened to the same result table. After
> finishing my analysis,  the content of the result table was saved to the
> specified path.

Umm... perhaps if I explain better what our "analysis" comprises.

We have an image of the seabed (actually hundreds).

This has two spots from laser lights to allow us to determine scale. We
measure these then cut a 50cmx50cm sub-image which we use for the actual
analysis. The "analyses" comprise selecting certaim brigtnes/color pixels
(manually- via the sliders, not pixel by pixel) to delineate areas of
rock, mud, sand, shell, gravel, cobbles, etc. Then we can get the area (&
%) of each image which is a particular substrate. The user then
marks/counts the numbers of various taxa which may be present; fish,
anemones, corals, urchins, starfish, etc, identified as well as the image
(& our knowledge of the taxa) permits.

Because we get several area results & counts for each image, simply
writing them to the results file doesn't let us know which record (row) in
the file pertains to which substrate or which count pertains to which taxa.

The scraping app I'm envisaging will allow extra fields to be entered (or
picked from a list) so that we can, for example, pick sand, then click
the result box to have a sand record grabbed & saved (with the extra field
"sand" included).

Because of the wide range of substrates & taxa which may or may not be
present, and issues with identifying them, we can't automate the process,
and we don't want users processing all the images for "sand" then
reopening & processing for each substrate, etc, as your approach (as I
read it) requires, we want an image at a time processed for all it's
content to be recorded.

This requires manual identification of each reading saved, which is not
currently possible. (If I'm wrong on this please tell me how!!)

My programming skills are close to zero, I can spell C but Gava is a
problem :-) So I'm totally dependent on someone to implement it for me...
hence the searching for $$ to fund it. I don't think such a a capability
is complex or difficult, but seems to be unusual.


Thanks for those who are offering suggestions, 'tis appreciated!


  Brent