quexf-discuss — queXF Discussion

Re: [Quexf-discuss] Several doubts about queXF usage

[M W <all...@gm...>]

I promise that this will be the last one, for today :-)

About the "Watch this directory ...." functionality:

I've introduced in the box of the directory something like
/home/user/quexf/doc/filled/ and clicked "Watch this directory"

and after a few minutes there is no "PDF-DONE" in that directory.

I've cleaned up the database and populated it again, with a fresh PDF
directory and now....

If I select manually "import pdf from directory" half of the PDF's
change to -DONE, but in the end I get an Internal Server Error
Warning. If, after that, I select watch this directory (recommended)
the remaining PDF's not previously recognized changed to -DONE.

When now reviewing the answers, they seemed well positioned (to the
right of the actual answers) (is that expected?), but quexf does not
guess any answer at all.

I wish my understanding of PHP and MYSQL was good enough to help you
out on the development of this tool :)

Rui Az.

Re: [Quexf-discuss] Several doubts about queXF usage

[M W <all...@gm...>]

Yet, another question:

* How can you completely remove a questionnaire from the system after
you have completed reviewing all the imported forms?

Re: [Quexf-discuss] Several doubts about queXF usage

[M W <all...@gm...>]

Hi Carlos,

Thank you for your help. I have now the filled in surveys and I found
the following:

On Fri, Dec 5, 2008 at 5:07 AM, Adam Zammit <az...@de...> wrote:
> Hello Carlos and Rui,
>
> Some of these issues come down to a design decision of storing forms imported in to queXF as Monochrome (1 bit) images.
>
> There are many advantages to doing this - such as very small file sizes at high resolutions and fast image operations.
>
> queXF was not designed to handle de-skewing or grayscale images because these can be fixed or converted using the scanning hardware or other open source software (such as imagemagick or ghostscript).

* In a survey with 500 pages at least in my case some pages suffer
from a slight inclination to the right or left of around 1º (not
always present). The Xerox machine does not do better than that. How
can you correct for this effect in such a large number of images using
imagemagick or ghostscript? Would it be possible, in theory, to
implement an algorithm which uses the vertical lines in barcode,
fitting the centers of the vertical strips as a function of page
height, to determine and correct for this small problems in the
scanning?

* When I am reviewing the forms, the guessed answers appear way to the
right and to the bottom of the actual answers in the screen. If I
click in "Centre page" now the guessed answers appear more or less
over the actual answers. I do not understand why this is happening.

* Only mouse clicks on the guessed answers and "Enter" seem to work.
For example,  down key scrolls down the page rather than moving the
cursor to the next choice... What am I missing? :)

* What do you think of the "bulk approve" all answers option  in the
review form?

Best.

Rui

Re: [Quexf-discuss] Several doubts about queXF usage

[Adam Z. <az...@de...>]

Hello Carlos and Rui,

Some of these issues come down to a design decision of storing forms imported in to queXF as Monochrome (1 bit) images.

There are many advantages to doing this - such as very small file sizes at high resolutions and fast image operations.

queXF was not designed to handle de-skewing or grayscale images because these can be fixed or converted using the scanning hardware or other open source software (such as imagemagick or ghostscript). 

The best way to import and add new forms to queXF are to scan them in using the following:

1. A4 paper size (although this can be modified now in new versions of queXF)
2. 300 DPI Resolution
3. Monochrome (1 bit)

To make it easier for people who wish to import the form without first scanning it - a new version of queXML was released today which allows a form to be generated with a white background. When this is imported as a monochrome image in to queXF - there are no difficulties with banding.

I still strongly recommend that if you wish to start a new form in queXF that you scan it in using the same scanner that you will be scanning completed forms in - use the settings above - and import in to queXF. If it does not import well, then the form should be scanned again with adjusted settings.

I intend to write some tools soon which will make it easier to modify the configuration of queXF by being able to graphically modify the pixel locations in the configuration file.

Regards,
Adam Zammit

Carlos Neira Cortizas wrote:
> Hi Adam, Rui,
> 
> That was the same problem I found few months ago (I have not tested the
> system again since that). I tried using the original PDF (output from
> QueXMLTools, test_original) instead of a scanned blank form (test_blank,
> in this case).
> 
> After many days testing the system I noticed then that QueXF is
> *very*sensitive to way this blank form is scanned. It depends not only
> on the configuration (DPI resolution, monochorme or grays) but on other
> hardware options (darker/lighter, dot density, entry angle of the forms,
> etc.) and maybe the scanner model too.
> 
> In my opinion it's absolutely necessary to touch advanced configuration
> parameters to get QueXF working the way Adam says, with no or few
> problems. Maybe it would be very useful for QueXF users a new GUI
> function (something like a test) to fix optimal settings in every
> scanner before starting.
> 
> 
> Adam Zammit escribió:
>> Dear Rui,
>>
>> It looks like you have used the form "test_original.pdf".
>>
>> At the moment: queXF does not handle this case well.
>>
>> Instead you should use the scanned form "test_blank.pdf"
>>
>> The idea is to band a form in the same format it will appear in your
>> scanning system. Scanned forms should be set to be 300DPI, Monochrome
>> (as in the test_blank.pdf example).
>>
>> You will not be able to create a new form in queXF using test_blank.pdf
>> if you have already imported "test_original.pdf" - as they have the same
>> page-id barcodes.
>>
>> If you do not have any important data in the queXF database - I
>> recommend you wipe it and start again.
>>
>> Regards,
>> Adam Zammit
>>
>>
>> M W wrote:
>>  
>>> Dear Adam,
>>>
>>> Thank you for your efforts.
>>>
>>> I guess something is seriously wrong in my installation although
>>> nothing odd is being reported!
>>>
>>> Attached follow a few screenshots more of the interactive banding
>>> result following the steps in the video and your previous
>>> instructions... Clearly not the desired result in any situation :)
>>>
>>> I just came across another problem ... Using the google cached file:
>>>  quexml.sourceforge.net/Examples/example.xml
>>>
>>> which corresponds more or less to the PDF example in the doc/
>>> directory and also, using an output from a survey I did in limesurvey,
>>> I went to https://dcarf.deakin.edu.au/quexmltools/ and produced a PDF
>>> file from the queXML using the online tool.
>>>
>>> The import of the PDF in queXF was successful... however, when doing
>>> the banding, the screen becomes white with nothing written except the
>>> left menu, after selecting the form. The last lines of the source code
>>> of the output page look like the following:
>>>
>>> ....
>>> <li><h3>Progress</h3>
>>> <ul><li><a href="javascript:link('mainobj','progress.php');">Display
>>> progress of form verification</a></li>
>>> <li><a href="javascript:link('mainobj','performance.php');">Display
>>> performance of verifiers (Completions per hour)</a></li></ul></li>
>>> <li><h3>Clients</h3>
>>> <ul><li><a href="javascript:link('mainobj','clients.php');">Add
>>> clients</a></li>
>>>
>>> <li><a
>>> href="javascript:link('mainobj','clientquestionnaire.php');">Assign
>>> clients to forms</a></li></ul></li>
>>> </ul>
>>> </div>
>>> <div id='main'><object class='embeddedobject' id='mainobj'
>>> data='new.php' standby='Loading panel...'
>>> type='application/xhtml+xml'><div>Error, try with
>>> Firefox</div></object></div>
>>>     </body>
>>>     </html>
>>>
>>> I am using Firefox 3 in Ubuntu.
>>>
>>> On Wed, Dec 3, 2008 at 1:13 AM, Adam Zammit <az...@de...>
>>> wrote:
>>>      
>>>> Dear Rui,
>>>>
>>>> It looks like the initial banding process is not working well for you -
>>>> as the entire box should be selected not just a small group.
>>>>
>>>> I have uploaded a video on interactive banding on the queXF website
>>>> here:
>>>>
>>>> http://quexf.sourceforge.net/node/24
>>>>
>>>> I think you should start the banding process again, firstly by deleting
>>>> all the current boxes that have been banded in those thin strips (use
>>>> interactive banding, right click on the boxes and select "Delete this
>>>> box group" - you can use the zoom feature if the boxes are too small to
>>>> access easily)
>>>>
>>>> Then, once all the existing box groups are deleted, band again using
>>>> the
>>>> same method as in the video - making sure that the boxes are selected
>>>> properly. The thin lines are not normal.
>>>>
>>>> Recognition will not improve with processing more forms. It is possible
>>>> to modify the recognition parameters in the config file, but I wouldn't
>>>> recommend it until you are comfortable with the system.
>>>>
>>>> When automatic recognition fails - the green box that should overlay
>>>> the
>>>> selected box will not appear (or appear over the wrong box).
>>>>
>>>> Good luck,
>>>> Adam Zammit
>>>>    
>>>> ------------------------------------------------------------------------
>>>>
>>>>
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>>
>>>>       
>>
>>  
>> ------------------------------------------------------------------------
>>
>> -------------------------------------------------------------------------
>> This SF.Net email is sponsored by the Moblin Your Move Developer's
>> challenge
>> Build the coolest Linux based applications with Moblin SDK & win great
>> prizes
>> Grand prize is a trip for two to an Open Source event anywhere in the
>> world
>> http://moblin-contest.org/redirect.php?banner_id=100&url=/
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Quexf-discuss mailing list
>> Que...@li...
>> https://lists.sourceforge.net/lists/listinfo/quexf-discuss
>>   
> 

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

Re: [Quexf-discuss] Several doubts about queXF usage

[M W <all...@gm...>]

Yes, you're right... now everything works in my linux server. Thanks a lot.

A few more comments:
* a windows XAMPP installation did not work well even when importing
test_blank.pdf. In the interactive banding, when you try to "draw" a
rectangle to select the boxes, while you have the left mouse button
clicked, appears a mouse cursor "forbidden", i.e., circle with one
line crossing the diameter, making impossible to do the banding;
* the operator does not have the ability to "bulk" approve all the
recognized answers of one form, which would save time;
* is it possible to delete completely from the database one form which
already has been processed fully or partially by the operators?
* more authentication mechanisms are desirable ... :)

Best regards.

Rui

On Wed, Dec 3, 2008 at 1:50 AM, Adam Zammit <az...@de...> wrote:
> Dear Rui,
>
> It looks like you have used the form "test_original.pdf".
>
> At the moment: queXF does not handle this case well.
>
> Instead you should use the scanned form "test_blank.pdf"
>
> The idea is to band a form in the same format it will appear in your
> scanning system. Scanned forms should be set to be 300DPI, Monochrome
> (as in the test_blank.pdf example).
>
> You will not be able to create a new form in queXF using test_blank.pdf
> if you have already imported "test_original.pdf" - as they have the same
> page-id barcodes.
>
> If you do not have any important data in the queXF database - I
> recommend you wipe it and start again.
>
> Regards,
> Adam Zammit
>
>
> M W wrote:
>> Dear Adam,
>>
>> Thank you for your efforts.
>>
>> I guess something is seriously wrong in my installation although
>> nothing odd is being reported!
>>
>> Attached follow a few screenshots more of the interactive banding
>> result following the steps in the video and your previous
>> instructions... Clearly not the desired result in any situation :)
>>
>> I just came across another problem ... Using the google cached file:
>>  quexml.sourceforge.net/Examples/example.xml
>>
>> which corresponds more or less to the PDF example in the doc/
>> directory and also, using an output from a survey I did in limesurvey,
>> I went to https://dcarf.deakin.edu.au/quexmltools/ and produced a PDF
>> file from the queXML using the online tool.
>>
>> The import of the PDF in queXF was successful... however, when doing
>> the banding, the screen becomes white with nothing written except the
>> left menu, after selecting the form. The last lines of the source code
>> of the output page look like the following:
>>
>> ....
>> <li><h3>Progress</h3>
>> <ul><li><a href="javascript:link('mainobj','progress.php');">Display
>> progress of form verification</a></li>
>> <li><a href="javascript:link('mainobj','performance.php');">Display
>> performance of verifiers (Completions per hour)</a></li></ul></li>
>> <li><h3>Clients</h3>
>> <ul><li><a href="javascript:link('mainobj','clients.php');">Add clients</a></li>
>>
>> <li><a href="javascript:link('mainobj','clientquestionnaire.php');">Assign
>> clients to forms</a></li></ul></li>
>> </ul>
>> </div>
>> <div id='main'><object class='embeddedobject' id='mainobj'
>> data='new.php' standby='Loading panel...'
>> type='application/xhtml+xml'><div>Error, try with
>> Firefox</div></object></div>
>>       </body>
>>       </html>
>>
>> I am using Firefox 3 in Ubuntu.
>>
>> On Wed, Dec 3, 2008 at 1:13 AM, Adam Zammit <az...@de...> wrote:
>>
>>> Dear Rui,
>>>
>>> It looks like the initial banding process is not working well for you -
>>> as the entire box should be selected not just a small group.
>>>
>>> I have uploaded a video on interactive banding on the queXF website here:
>>>
>>> http://quexf.sourceforge.net/node/24
>>>
>>> I think you should start the banding process again, firstly by deleting
>>> all the current boxes that have been banded in those thin strips (use
>>> interactive banding, right click on the boxes and select "Delete this
>>> box group" - you can use the zoom feature if the boxes are too small to
>>> access easily)
>>>
>>> Then, once all the existing box groups are deleted, band again using the
>>> same method as in the video - making sure that the boxes are selected
>>> properly. The thin lines are not normal.
>>>
>>> Recognition will not improve with processing more forms. It is possible
>>> to modify the recognition parameters in the config file, but I wouldn't
>>> recommend it until you are comfortable with the system.
>>>
>>> When automatic recognition fails - the green box that should overlay the
>>> selected box will not appear (or appear over the wrong box).
>>>
>>> Good luck,
>>> Adam Zammit
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>
> --
> Adam Zammit, Research Fellow in Survey Technology
>
> Deakin University 221 Burwood Hwy, Burwood Australia.
> Phone: 03 9251 7290 International: +61 3 9251 7290
> Fax: 03 9251 7632 International: +61 3 9251 7632
> Email: ada...@de...
> Website: http://www.deakin.edu.au/buslaw/dcarf/
> Deakin University CRICOS Provider Code 00113B (Vic)
>
> Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.
>
> Deakin University does not warrant that this email and any attachments are error or virus free.
>
>
>

Re: [Quexf-discuss] Several doubts about queXF usage

[Adam Z. <az...@de...>]

Dear Rui,

It looks like you have used the form "test_original.pdf".

At the moment: queXF does not handle this case well.

Instead you should use the scanned form "test_blank.pdf"

The idea is to band a form in the same format it will appear in your
scanning system. Scanned forms should be set to be 300DPI, Monochrome
(as in the test_blank.pdf example).

You will not be able to create a new form in queXF using test_blank.pdf
if you have already imported "test_original.pdf" - as they have the same
page-id barcodes.

If you do not have any important data in the queXF database - I
recommend you wipe it and start again.

Regards,
Adam Zammit


M W wrote:
> Dear Adam,
>
> Thank you for your efforts.
>
> I guess something is seriously wrong in my installation although
> nothing odd is being reported!
>
> Attached follow a few screenshots more of the interactive banding
> result following the steps in the video and your previous
> instructions... Clearly not the desired result in any situation :)
>
> I just came across another problem ... Using the google cached file:
>  quexml.sourceforge.net/Examples/example.xml
>
> which corresponds more or less to the PDF example in the doc/
> directory and also, using an output from a survey I did in limesurvey,
> I went to https://dcarf.deakin.edu.au/quexmltools/ and produced a PDF
> file from the queXML using the online tool.
>
> The import of the PDF in queXF was successful... however, when doing
> the banding, the screen becomes white with nothing written except the
> left menu, after selecting the form. The last lines of the source code
> of the output page look like the following:
>
> ....
> <li><h3>Progress</h3>
> <ul><li><a href="javascript:link('mainobj','progress.php');">Display
> progress of form verification</a></li>
> <li><a href="javascript:link('mainobj','performance.php');">Display
> performance of verifiers (Completions per hour)</a></li></ul></li>
> <li><h3>Clients</h3>
> <ul><li><a href="javascript:link('mainobj','clients.php');">Add clients</a></li>
>
> <li><a href="javascript:link('mainobj','clientquestionnaire.php');">Assign
> clients to forms</a></li></ul></li>
> </ul>
> </div>
> <div id='main'><object class='embeddedobject' id='mainobj'
> data='new.php' standby='Loading panel...'
> type='application/xhtml+xml'><div>Error, try with
> Firefox</div></object></div>
> 	</body>
> 	</html>
>
> I am using Firefox 3 in Ubuntu.
>
> On Wed, Dec 3, 2008 at 1:13 AM, Adam Zammit <az...@de...> wrote:
>   
>> Dear Rui,
>>
>> It looks like the initial banding process is not working well for you -
>> as the entire box should be selected not just a small group.
>>
>> I have uploaded a video on interactive banding on the queXF website here:
>>
>> http://quexf.sourceforge.net/node/24
>>
>> I think you should start the banding process again, firstly by deleting
>> all the current boxes that have been banded in those thin strips (use
>> interactive banding, right click on the boxes and select "Delete this
>> box group" - you can use the zoom feature if the boxes are too small to
>> access easily)
>>
>> Then, once all the existing box groups are deleted, band again using the
>> same method as in the video - making sure that the boxes are selected
>> properly. The thin lines are not normal.
>>
>> Recognition will not improve with processing more forms. It is possible
>> to modify the recognition parameters in the config file, but I wouldn't
>> recommend it until you are comfortable with the system.
>>
>> When automatic recognition fails - the green box that should overlay the
>> selected box will not appear (or appear over the wrong box).
>>
>> Good luck,
>> Adam Zammit
>>     
>>
>> ------------------------------------------------------------------------
>>
>>
>> ------------------------------------------------------------------------
>>

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

Re: [Quexf-discuss] Several doubts about queXF usage

[M W <all...@gm...>]

Dear Adam,

Thank you for your efforts.

I guess something is seriously wrong in my installation although
nothing odd is being reported!

Attached follow a few screenshots more of the interactive banding
result following the steps in the video and your previous
instructions... Clearly not the desired result in any situation :)

I just came across another problem ... Using the google cached file:
 quexml.sourceforge.net/Examples/example.xml

which corresponds more or less to the PDF example in the doc/
directory and also, using an output from a survey I did in limesurvey,
I went to https://dcarf.deakin.edu.au/quexmltools/ and produced a PDF
file from the queXML using the online tool.

The import of the PDF in queXF was successful... however, when doing
the banding, the screen becomes white with nothing written except the
left menu, after selecting the form. The last lines of the source code
of the output page look like the following:

....
<li><h3>Progress</h3>
<ul><li><a href="javascript:link('mainobj','progress.php');">Display
progress of form verification</a></li>
<li><a href="javascript:link('mainobj','performance.php');">Display
performance of verifiers (Completions per hour)</a></li></ul></li>
<li><h3>Clients</h3>
<ul><li><a href="javascript:link('mainobj','clients.php');">Add clients</a></li>

<li><a href="javascript:link('mainobj','clientquestionnaire.php');">Assign
clients to forms</a></li></ul></li>
</ul>
</div>
<div id='main'><object class='embeddedobject' id='mainobj'
data='new.php' standby='Loading panel...'
type='application/xhtml+xml'><div>Error, try with
Firefox</div></object></div>
	</body>
	</html>

I am using Firefox 3 in Ubuntu.

On Wed, Dec 3, 2008 at 1:13 AM, Adam Zammit <az...@de...> wrote:
>
> Dear Rui,
>
> It looks like the initial banding process is not working well for you -
> as the entire box should be selected not just a small group.
>
> I have uploaded a video on interactive banding on the queXF website here:
>
> http://quexf.sourceforge.net/node/24
>
> I think you should start the banding process again, firstly by deleting
> all the current boxes that have been banded in those thin strips (use
> interactive banding, right click on the boxes and select "Delete this
> box group" - you can use the zoom feature if the boxes are too small to
> access easily)
>
> Then, once all the existing box groups are deleted, band again using the
> same method as in the video - making sure that the boxes are selected
> properly. The thin lines are not normal.
>
> Recognition will not improve with processing more forms. It is possible
> to modify the recognition parameters in the config file, but I wouldn't
> recommend it until you are comfortable with the system.
>
> When automatic recognition fails - the green box that should overlay the
> selected box will not appear (or appear over the wrong box).
>
> Good luck,
> Adam Zammit

Re: [Quexf-discuss] Several doubts about queXF usage

[Adam Z. <az...@de...>]

Dear Rui,

It looks like the initial banding process is not working well for you -
as the entire box should be selected not just a small group.

I have uploaded a video on interactive banding on the queXF website here:

http://quexf.sourceforge.net/node/24

I think you should start the banding process again, firstly by deleting
all the current boxes that have been banded in those thin strips (use
interactive banding, right click on the boxes and select "Delete this
box group" - you can use the zoom feature if the boxes are too small to
access easily)

Then, once all the existing box groups are deleted, band again using the
same method as in the video - making sure that the boxes are selected
properly. The thin lines are not normal.

Recognition will not improve with processing more forms. It is possible
to modify the recognition parameters in the config file, but I wouldn't
recommend it until you are comfortable with the system.

When automatic recognition fails - the green box that should overlay the
selected box will not appear (or appear over the wrong box).

Good luck,
Adam Zammit



M W wrote:
> Thanks again for your help. After reimporting the filled in forms I
> was able to finally see some kind of automatic detection happening in
> the review process... however, a few questions appeared :)
>
> As you can see in the screenshot attached, the "automatic detection"
> only shows a very narrow green vertical strip over the box, which is
> very hard to spot visually... is this a normal behavior? I mean, it is
> very difficult to understand if the detection was well made or not at
> first sight.
>
> In addition, you said that a small proportion of boxes will not be
> detected automatically... in my case, most of them are not detected
> correctly using the sample in the doc/ directory. I did not look into
> the code but ... does the recognition improves as we process more and
> more forms? I am not using tesseract if that matters.
>
> Finally, should I expect some sort of visual/color indication when the
> automatic recognition fails or just an absence of a green vertical strip?
>
> Best,
>
> Rui
>
>
>     Also - a small proportion of boxes will never be detected
>     automatically
>     by queXF  - this is the reason for the verification step. The
>     operators
>     will then select the correct box by clicking on it, or pressing the
>     applicable number key on the keyboard, as described in the
>     documentation: http://quexf.sourceforge.net/node/16 under
>     "Verification"
>
>     Regards,
>     Adam Zammit
>
>     M W wrote:
>     > Dear Adam,
>     >
>     > Thanks a lot for your help. I've been able to do the interactive
>     > banding correctly now. The thing was, since it takes a little bit of
>     > time for the region to turn red (at least in my computer), I did not
>     > realize that one left click was enough. Another source of
>     confusion is
>     > that in the screenshots available in queXF website, only the
>     boxes are
>     > green... in my case, all the selected region is green/red but the
>     > boxes have vertical stripes and have a lighter colour.
>     >
>     > Now, the operator review still does not detect which box has a cross
>     > in it... everything remains orange with no special highlight :/
>     > Attached follows a screen capture. In the screenshot it appears that
>     > the light orange is slightly below the real boxes.... (in the
>     > screenshot, the 3 boxes of the second question are in analysis,
>     but a
>     > similar situation happens in the others after pressing Enter)
>     >
>     > Thank you! Hopefully, all my institute will be using queXF for
>     quality
>     > assessment as soon as I get it working well.
>     >
>     > Best.
>     >
>     > Rui
>     >
>     >
>     > On 12/1/08, *Adam Zammit* <az...@de...
>     <mailto:az...@de...>
>     > <mailto:az...@de... <mailto:az...@de...>>>
>     wrote:
>     >
>     >     Dear Rui,
>     >
>     >     To band (interactively) the boxes to the right of "The first
>     part of a
>     >     matrix question" you would do the following:
>     >
>     >     1. Click and drag from the top left of the 3 boxes to the
>     bottom right
>     >     of the 3 boxes so that you have selected an area surrounding
>     all 3
>     >     boxes
>     >     (but not much more)
>     >     2. Click once within this selected area
>     >     3. The 3 boxes should then be detected and highlighted in red
>     >     4. Right click on one of the boxes, and select "Single choice"
>     >     5. The boxes should then be highlighted as green green
>     >
>     >     By left clicking on the boxes - you are bringing up the
>     screen which
>     >     allows you to assign a variable name to the boxes.
>     >
>     >     Currently, the 'width' option is only supported for
>     barcodes, and is
>     >     automatically entered. This box can be safely ignored.
>     >
>     >     If you want the variable name to appear in your data output
>     (such
>     >     as the
>     >     DDI or CSV file) - enter a variable name in this screen.
>     >
>     >     The screen also shows the boxes highlighted as an image at
>     the top to
>     >     confirm that the boxes selected are correct.
>     >
>     >
>     >     When banding, if you click in the selected area more than once,
>     >     then the
>     >     boxes will be detected more than once. This will make the
>     boxes appear
>     >     "darker" as there are more red boxes being overlaid over
>     each other.
>     >
>     >     When banding, the idea is to select boxes in a group - such
>     as the 3
>     >     boxes to the right of "The first part of a matrix question"
>     would
>     >     all be
>     >     selected together. Then, for example, they will be given the
>     variable
>     >     name "A1a". The 3 boxes to the right of "The second question
>     in a
>     >     matrix
>     >     question" would be selected as a group and named, for
>     example, "A1b"
>     >
>     >     It is unclear what the problem is with the operator
>     verification: I
>     >     think it would stem back to the banding process.
>     >
>     >     It may be easier to start off working with the original banding
>     >     process
>     >     (Where you click in the top left corner of the group of
>     boxes, then
>     >     click in the bottom right corner of the group of boxes to
>     select a box
>     >     group) if you are having trouble with the interactive process.
>     >
>     >     Regards,
>     >     Adam Zammit
>     >
>     >
>     >     M W wrote:
>     >     > Hi.
>     >     >
>     >     > I am trying to understand how queXF works but I am having
>     several
>     >     > doubts. I am using the latest version, the PDF sample from
>     the doc/
>     >     > directory, Firefox 3.
>     >     >
>     >     > Let us suppose that I want to "Band Form" interactively ,
>     >     specifically
>     >     > the boxes to the right of "The first part of a matrix
>     question",
>     >     "The
>     >     > Second question in a matrix question" (page 2). I do a row
>     at each
>     >     > time, i.e., I select a region that encloses three boxes to
>     the right
>     >     > of the "The first part of a matrix question". Then I have
>     to click a
>     >     > few times around in order to get the region red such that
>     I can
>     >     click
>     >     > it. When the red appears, I finally click and I am brought
>     to a
>     >     dialog
>     >     > which still has the three boxes all red. Is this normal?
>     >     shouldn't the
>     >     > program have recgonized by now that there are three boxes
>     and mark
>     >     > them in color separately with white between? I tried to
>     delete space
>     >     > between boxes and what happens is that the program changes
>     all the
>     >     > color in red to something lighter... in all the region,
>     including
>     >     > boxes and the space between the boxes!
>     >     >
>     >     > Btw, what is the width field? Should I assign the same
>     name to the
>     >     > three boxes such they can be recognized as a group?
>     >     >
>     >     > Also, if I select only one box in order to continue the
>     >     workflow, when
>     >     > the operator is reviewing, I always see everything orange and
>     >     nothing
>     >     > green as if none box was marked with a cross....
>     >     >
>     >     > What could be possibly wrong?
>     >     >
>     >     > Thanks in advance.
>     >     >
>     >     > Rui
>     >
>     >     >
>     >    
>     ------------------------------------------------------------------------
>     >     >
>     >     >
>     >    
>     -------------------------------------------------------------------------
>     >     > This SF.Net email is sponsored by the Moblin Your Move
>     >     Developer's challenge
>     >     > Build the coolest Linux based applications with Moblin SDK
>     & win
>     >     great prizes
>     >     > Grand prize is a trip for two to an Open Source event anywhere
>     >     in the world
>     >     > http://moblin-contest.org/redirect.php?banner_id=100&url=/
>     <http://moblin-contest.org/redirect.php?banner_id=100&url=/>
>     >     <http://moblin-contest.org/redirect.php?banner_id=100&url=/
>     <http://moblin-contest.org/redirect.php?banner_id=100&url=/>>
>     >     >
>     >    
>     ------------------------------------------------------------------------
>     >     >
>     >     > _______________________________________________
>     >     > Quexf-discuss mailing list
>     >     > Que...@li...
>     <mailto:Que...@li...>
>     >     <mailto:Que...@li...
>     <mailto:Que...@li...>>
>     >     > https://lists.sourceforge.net/lists/listinfo/quexf-discuss
>     >     >
>     >
>     >     --
>     >     Adam Zammit, Research Fellow in Survey Technology
>     >
>     >     Deakin University 221 Burwood Hwy, Burwood Australia.
>     >     Phone: 03 9251 7290 International: +61 3 9251 7290
>     >     Fax: 03 9251 7632 International: +61 3 9251 7632
>     >     Email: ada...@de...
>     <mailto:ada...@de...>
>     <mailto:ada...@de... <mailto:ada...@de...>>
>     >     Website: http://www.deakin.edu.au/buslaw/dcarf/
>     >     Deakin University CRICOS Provider Code 00113B (Vic)
>     >
>     >     Important Notice: The contents of this email are intended solely
>     >     for the named addressee and are confidential; any unauthorised
>     >     use, reproduction or storage of the contents is expressly
>     >     prohibited. If you have received this email in error, please
>     >     delete it and any attachments immediately and advise the
>     sender by
>     >     return email or telephone.
>     >
>     >     Deakin University does not warrant that this email and any
>     >     attachments are error or virus free.
>     >
>     >
>     >
>     >
>     >
>     >
>     ------------------------------------------------------------------------
>     >
>
>     --
>     Adam Zammit, Research Fellow in Survey Technology
>
>     Deakin University 221 Burwood Hwy, Burwood Australia.
>     Phone: 03 9251 7290 International: +61 3 9251 7290
>     Fax: 03 9251 7632 International: +61 3 9251 7632
>     Email: ada...@de... <mailto:ada...@de...>
>     Website: http://www.deakin.edu.au/buslaw/dcarf/
>     Deakin University CRICOS Provider Code 00113B (Vic)
>
>     Important Notice: The contents of this email are intended solely
>     for the named addressee and are confidential; any unauthorised
>     use, reproduction or storage of the contents is expressly
>     prohibited. If you have received this email in error, please
>     delete it and any attachments immediately and advise the sender by
>     return email or telephone.
>
>     Deakin University does not warrant that this email and any
>     attachments are error or virus free.
>
>
>
>
> ------------------------------------------------------------------------
>

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

Re: [Quexf-discuss] Several doubts about queXF usage

[Adam Z. <az...@de...>]

Dear Rui,

Once the banding has been done correctly, for the boxes to be detected
by the system, the forms need to be imported again.

Only the forms imported (using: Import a directory of PDF files tool)
after banding is correct will have the boxes detected correctly.

If after importing the forms again the boxes are still not detected,
please let me know as it may be a different problem.

Also - a small proportion of boxes will never be detected automatically
by queXF  - this is the reason for the verification step. The operators
will then select the correct box by clicking on it, or pressing the
applicable number key on the keyboard, as described in the
documentation: http://quexf.sourceforge.net/node/16 under "Verification"

Regards,
Adam Zammit

M W wrote:
> Dear Adam,
>
> Thanks a lot for your help. I've been able to do the interactive
> banding correctly now. The thing was, since it takes a little bit of
> time for the region to turn red (at least in my computer), I did not
> realize that one left click was enough. Another source of confusion is
> that in the screenshots available in queXF website, only the boxes are
> green... in my case, all the selected region is green/red but the
> boxes have vertical stripes and have a lighter colour.
>
> Now, the operator review still does not detect which box has a cross
> in it... everything remains orange with no special highlight :/
> Attached follows a screen capture. In the screenshot it appears that
> the light orange is slightly below the real boxes.... (in the
> screenshot, the 3 boxes of the second question are in analysis, but a
> similar situation happens in the others after pressing Enter)
>
> Thank you! Hopefully, all my institute will be using queXF for quality
> assessment as soon as I get it working well.
>
> Best.
>
> Rui
>
>
> On 12/1/08, *Adam Zammit* <az...@de...
> <mailto:az...@de...>> wrote:
>
>     Dear Rui,
>
>     To band (interactively) the boxes to the right of "The first part of a
>     matrix question" you would do the following:
>
>     1. Click and drag from the top left of the 3 boxes to the bottom right
>     of the 3 boxes so that you have selected an area surrounding all 3
>     boxes
>     (but not much more)
>     2. Click once within this selected area
>     3. The 3 boxes should then be detected and highlighted in red
>     4. Right click on one of the boxes, and select "Single choice"
>     5. The boxes should then be highlighted as green green
>
>     By left clicking on the boxes - you are bringing up the screen which
>     allows you to assign a variable name to the boxes.
>
>     Currently, the 'width' option is only supported for barcodes, and is
>     automatically entered. This box can be safely ignored.
>
>     If you want the variable name to appear in your data output (such
>     as the
>     DDI or CSV file) - enter a variable name in this screen.
>
>     The screen also shows the boxes highlighted as an image at the top to
>     confirm that the boxes selected are correct.
>
>
>     When banding, if you click in the selected area more than once,
>     then the
>     boxes will be detected more than once. This will make the boxes appear
>     "darker" as there are more red boxes being overlaid over each other.
>
>     When banding, the idea is to select boxes in a group - such as the 3
>     boxes to the right of "The first part of a matrix question" would
>     all be
>     selected together. Then, for example, they will be given the variable
>     name "A1a". The 3 boxes to the right of "The second question in a
>     matrix
>     question" would be selected as a group and named, for example, "A1b"
>
>     It is unclear what the problem is with the operator verification: I
>     think it would stem back to the banding process.
>
>     It may be easier to start off working with the original banding
>     process
>     (Where you click in the top left corner of the group of boxes, then
>     click in the bottom right corner of the group of boxes to select a box
>     group) if you are having trouble with the interactive process.
>
>     Regards,
>     Adam Zammit
>
>
>     M W wrote:
>     > Hi.
>     >
>     > I am trying to understand how queXF works but I am having several
>     > doubts. I am using the latest version, the PDF sample from the doc/
>     > directory, Firefox 3.
>     >
>     > Let us suppose that I want to "Band Form" interactively ,
>     specifically
>     > the boxes to the right of "The first part of a matrix question",
>     "The
>     > Second question in a matrix question" (page 2). I do a row at each
>     > time, i.e., I select a region that encloses three boxes to the right
>     > of the "The first part of a matrix question". Then I have to click a
>     > few times around in order to get the region red such that I can
>     click
>     > it. When the red appears, I finally click and I am brought to a
>     dialog
>     > which still has the three boxes all red. Is this normal?
>     shouldn't the
>     > program have recgonized by now that there are three boxes and mark
>     > them in color separately with white between? I tried to delete space
>     > between boxes and what happens is that the program changes all the
>     > color in red to something lighter... in all the region, including
>     > boxes and the space between the boxes!
>     >
>     > Btw, what is the width field? Should I assign the same name to the
>     > three boxes such they can be recognized as a group?
>     >
>     > Also, if I select only one box in order to continue the
>     workflow, when
>     > the operator is reviewing, I always see everything orange and
>     nothing
>     > green as if none box was marked with a cross....
>     >
>     > What could be possibly wrong?
>     >
>     > Thanks in advance.
>     >
>     > Rui
>
>     >
>     ------------------------------------------------------------------------
>     >
>     >
>     -------------------------------------------------------------------------
>     > This SF.Net email is sponsored by the Moblin Your Move
>     Developer's challenge
>     > Build the coolest Linux based applications with Moblin SDK & win
>     great prizes
>     > Grand prize is a trip for two to an Open Source event anywhere
>     in the world
>     > http://moblin-contest.org/redirect.php?banner_id=100&url=/
>     <http://moblin-contest.org/redirect.php?banner_id=100&url=/>
>     >
>     ------------------------------------------------------------------------
>     >
>     > _______________________________________________
>     > Quexf-discuss mailing list
>     > Que...@li...
>     <mailto:Que...@li...>
>     > https://lists.sourceforge.net/lists/listinfo/quexf-discuss
>     >
>
>     --
>     Adam Zammit, Research Fellow in Survey Technology
>
>     Deakin University 221 Burwood Hwy, Burwood Australia.
>     Phone: 03 9251 7290 International: +61 3 9251 7290
>     Fax: 03 9251 7632 International: +61 3 9251 7632
>     Email: ada...@de... <mailto:ada...@de...>
>     Website: http://www.deakin.edu.au/buslaw/dcarf/
>     Deakin University CRICOS Provider Code 00113B (Vic)
>
>     Important Notice: The contents of this email are intended solely
>     for the named addressee and are confidential; any unauthorised
>     use, reproduction or storage of the contents is expressly
>     prohibited. If you have received this email in error, please
>     delete it and any attachments immediately and advise the sender by
>     return email or telephone.
>
>     Deakin University does not warrant that this email and any
>     attachments are error or virus free.
>
>
>
>
>
> ------------------------------------------------------------------------
>

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

Re: [Quexf-discuss] Several doubts about queXF usage

[Adam Z. <az...@de...>]

Dear Rui,

To band (interactively) the boxes to the right of "The first part of a
matrix question" you would do the following:

1. Click and drag from the top left of the 3 boxes to the bottom right
of the 3 boxes so that you have selected an area surrounding all 3 boxes
(but not much more)
2. Click once within this selected area
3. The 3 boxes should then be detected and highlighted in red
4. Right click on one of the boxes, and select "Single choice"
5. The boxes should then be highlighted as green green

By left clicking on the boxes - you are bringing up the screen which
allows you to assign a variable name to the boxes.

Currently, the 'width' option is only supported for barcodes, and is
automatically entered. This box can be safely ignored.

If you want the variable name to appear in your data output (such as the
DDI or CSV file) - enter a variable name in this screen.

The screen also shows the boxes highlighted as an image at the top to
confirm that the boxes selected are correct.


When banding, if you click in the selected area more than once, then the
boxes will be detected more than once. This will make the boxes appear
"darker" as there are more red boxes being overlaid over each other.

When banding, the idea is to select boxes in a group - such as the 3
boxes to the right of "The first part of a matrix question" would all be
selected together. Then, for example, they will be given the variable
name "A1a". The 3 boxes to the right of "The second question in a matrix
question" would be selected as a group and named, for example, "A1b"

It is unclear what the problem is with the operator verification: I
think it would stem back to the banding process.

It may be easier to start off working with the original banding process
(Where you click in the top left corner of the group of boxes, then
click in the bottom right corner of the group of boxes to select a box
group) if you are having trouble with the interactive process.

Regards,
Adam Zammit

M W wrote:
> Hi.
>
> I am trying to understand how queXF works but I am having several
> doubts. I am using the latest version, the PDF sample from the doc/
> directory, Firefox 3.
>
> Let us suppose that I want to "Band Form" interactively , specifically
> the boxes to the right of "The first part of a matrix question", "The
> Second question in a matrix question" (page 2). I do a row at each
> time, i.e., I select a region that encloses three boxes to the right
> of the "The first part of a matrix question". Then I have to click a
> few times around in order to get the region red such that I can click
> it. When the red appears, I finally click and I am brought to a dialog
> which still has the three boxes all red. Is this normal? shouldn't the
> program have recgonized by now that there are three boxes and mark
> them in color separately with white between? I tried to delete space
> between boxes and what happens is that the program changes all the
> color in red to something lighter... in all the region, including
> boxes and the space between the boxes!
>
> Btw, what is the width field? Should I assign the same name to the
> three boxes such they can be recognized as a group?
>
> Also, if I select only one box in order to continue the workflow, when
> the operator is reviewing, I always see everything orange and nothing
> green as if none box was marked with a cross....
>
> What could be possibly wrong?
>
> Thanks in advance.
>
> Rui
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> ------------------------------------------------------------------------
>
> _______________________________________________
> Quexf-discuss mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quexf-discuss
>   

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

[Quexf-discuss] Several doubts about queXF usage

[M W <all...@gm...>]

Hi.

I am trying to understand how queXF works but I am having several doubts. I
am using the latest version, the PDF sample from the doc/ directory, Firefox
3.

Let us suppose that I want to "Band Form" interactively , specifically the
boxes to the right of "The first part of a matrix question", "The Second
question in a matrix question" (page 2). I do a row at each time, i.e., I
select a region that encloses three boxes to the right of the "The first
part of a matrix question". Then I have to click a few times around in order
to get the region red such that I can click it. When the red appears, I
finally click and I am brought to a dialog which still has the three boxes
all red. Is this normal? shouldn't the program have recgonized by now that
there are three boxes and mark them in color separately with white between?
I tried to delete space between boxes and what happens is that the program
changes all the color in red to something lighter... in all the region,
including boxes and the space between the boxes!

Btw, what is the width field? Should I assign the same name to the three
boxes such they can be recognized as a group?

Also, if I select only one box in order to continue the workflow, when the
operator is reviewing, I always see everything orange and nothing green as
if none box was marked with a cross....

What could be possibly wrong?

Thanks in advance.

Rui

[Quexf-discuss] queXF 1.5.0 Released

[Adam Z. <az...@de...>]

Hello All,

queXF 1.5.0 has been released today.

See here:

http://quexf.sourceforge.net/node/23

For new features, bug fixes, and how to upgrade from 1.4.0

Regards,
Adam Zammit

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

[Quexf-discuss] queXF 1.4.0 Released

[Adam Z. <az...@de...>]

Hello all,

queXF 1.4.0 has been released.

New features include:
Ability to read Codabar barcodes
"Client view" of forms: Review function for clients where they can only
review forms assigned to them

For more details including how to upgrade, see here:

http://quexf.sourceforge.net/node/22

Regards,
Adam Zammit

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

Re: [Quexf-discuss] Need some help setting up 1.3.1 to use 8.5 x 11 paper

[Adam Z. <az...@de...>]

Hi Nate,

It is good to hear it is working.

For future reference: for someone who wants to modify the new
configuration file (in queXF 1.3.0 and above) for their page size - here
are the steps to follow:

1. Use ghostscript to convert the PDF file in to PNG's at a resolution
of 300DPI. You can use this example command line:

gs -sDEVICE=pngmono -r300 -sOutputFile=TEMP%d.png -dNOPAUSE -dBATCH
filename.pdf

Where 'filename.pdf ' is the pdf to convert. After running this command,
you will have as many png's are there are pages in the PDF each called
TEMP1.png, TEMP2.png...

2. Open this up in a graphics editor (such as the GIMP) and then fill
out the details in the configuration file.

Most configuration options require you to estimate a "box" (the top left
x and y coordinates and the bottom right x and y coordinates) around an
element such as the barcode or page guides.

Regards,
Adam Zammit

Nate Straight wrote:
> Actually, I think I figured it out. The horizontal banding seems to be
> working now, as well. Wonderful.
>
> -Nate
>
> On Fri, Nov 14, 2008 at 3:30 PM, Nate Straight <nat...@gm...
> <mailto:nat...@gm...>> wrote:
>
>     Hi,
>
>     I see some new functions in the update that would be very useful
>     to me, and I just had a problem earlier today trying to band some
>     horizontal box groups (still can't get this to work on my old
>     build), so I figured I'd give it a shot. This version sure
>     installs easier and with less setup than the prior version.
>
>     I'm running into the same problem that I posted about a few weeks
>     ago, of having to scan in 8.5 x 11 surveys rather than A4. I'm not
>     sure what resolution I need to load my PDFs at into GIMP or some
>     other image editor to find the right pixel location of the various
>     guides, or the correct page size.
>
>     I'm attaching a stupid blank survey that's been printed and
>     scanned in at 8.5 x 11 size. I was able to scan and band this
>     survey properly on the build I had before, but it won't band
>     correctly now (though it will scan in and identify the barcode
>     correctly, just not the boxes). Any help would be great.
>
>     -Nate
>
>
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> ------------------------------------------------------------------------
>
> _______________________________________________
> Quexf-discuss mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quexf-discuss
>   

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

Re: [Quexf-discuss] Need some help setting up 1.3.1 to use 8.5 x 11 paper

[Nate S. <nat...@gm...>]

Actually, I think I figured it out. The horizontal banding seems to be
working now, as well. Wonderful.

-Nate

On Fri, Nov 14, 2008 at 3:30 PM, Nate Straight <nat...@gm...>wrote:

> Hi,
>
> I see some new functions in the update that would be very useful to me, and
> I just had a problem earlier today trying to band some horizontal box groups
> (still can't get this to work on my old build), so I figured I'd give it a
> shot. This version sure installs easier and with less setup than the prior
> version.
>
> I'm running into the same problem that I posted about a few weeks ago, of
> having to scan in 8.5 x 11 surveys rather than A4. I'm not sure what
> resolution I need to load my PDFs at into GIMP or some other image editor to
> find the right pixel location of the various guides, or the correct page
> size.
>
> I'm attaching a stupid blank survey that's been printed and scanned in at
> 8.5 x 11 size. I was able to scan and band this survey properly on the build
> I had before, but it won't band correctly now (though it will scan in and
> identify the barcode correctly, just not the boxes). Any help would be
> great.
>
> -Nate
>

[Quexf-discuss] queXF 1.3.1 (Bug fix release)

[Adam Z. <az...@de...>]

Hello all,

queXF 1.3.1 has been released today.

It fixes a bug in the new banding system which caused scripts in the
browser to run very slowly - making banding difficult.

If you have just installed 1.3.0, you can download 1.3.1 and update the
file admin/bandajax.php (the only change)

See here for details:

http://quexf.sourceforge.net/node/21

Regards,
Adam Zammit

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

[Quexf-discuss] queXF 1.3.0 released

[Adam Z. <az...@de...>]

Hello everyone,

queXF 1.3.0 has been released today (queXF 1.3.0 windows executable
coming soon)

queXF 1.3.0 has the following new features:

* Interactive banding of forms (faster - using AJAX - thanks to
Prototype JS and Rectangular Marquee Tool)
* More accurate barcode detection
* All pages XHTML 1.1 Compliant
* Page size modifications possible in config.inc.php
* Banding zooming works without resetting page

See here: http://quexf.sourceforge.net/node/20
for full details.

I would also appreciate if anyone who is using non-A4 page sizes (the
default in queXF) could test the new settings in config.inc.php. I would
be more than willing to help with configuration issues.

Regards,
Adam Zammit

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

Re: [Quexf-discuss] Attempt to use OCR returns Fatal Error "Call to undefined function ST_Guess() in functions.import.php on line 102"

[Adam Z. <az...@de...>]

Dear Nate,

Your build of queXF is based on the SVN trunk (latest, but unstable)
build of queXF. This build has a (very) experimental OCR module
installed instead of the Tesseract option in the stable build of queXF.
It is unlikely to work well without a bit of modification as it is still
in development.

I have not had great success with the Tesseract module either. It tends
to be only about 50-60% accurate (I'm sure it can me much better but
needs some development work). We have found that using OCR with a low
rate of success takes longer for the operator to verify than without any
OCR at all (as characters have to be deleted/replaced).

I recommend that unless you are interested in spending some time
developing the interface between queXF and Tesseract - and training
Tesseract - that you disable OCR.

Regards,
Adam Zammit

Nate Straight wrote:
> Hello. Thanks to Adam's diligence, I have a working build of QueXF
> that can accept US 8.5x11 instead of international A4 scanned-in surveys.
>
> It's working great for standard surveys with just checked-off boxes.
> I'm having trouble incorporating the experimental OCR module.
>
> I have installed Tesseract and ImageMagick, and pointed my
> config.inc.php to their binaries. The admin/testconfig.php file
> confirms both.
>
> When I try to upload a directory of filled-out surveys with text /
> number fields, I get the error described in the subject heading, namely:
>
> "Fatal error: Call to undefined function ST_Guess() in
> *pathname*/functions.import.php on line 102"
>
> Is this an ImageMagick or Tesseract function that I should be looking
> for, or is it an internal function to QueXF? Is it even a function?
>
> Thanks in advance. Looking forward to increasing the processing power
> of this already great piece of software.
>
> -Nate
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> ------------------------------------------------------------------------
>
> _______________________________________________
> Quexf-discuss mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quexf-discuss
>   

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

[Quexf-discuss] Attempt to use OCR returns Fatal Error "Call to undefined function ST_Guess() in functions.import.php on line 102"

[Nate S. <nat...@gm...>]

Hello. Thanks to Adam's diligence, I have a working build of QueXF that can
accept US 8.5x11 instead of international A4 scanned-in surveys.

It's working great for standard surveys with just checked-off boxes. I'm
having trouble incorporating the experimental OCR module.

I have installed Tesseract and ImageMagick, and pointed my config.inc.php to
their binaries. The admin/testconfig.php file confirms both.

When I try to upload a directory of filled-out surveys with text / number
fields, I get the error described in the subject heading, namely:

"Fatal error: Call to undefined function ST_Guess() in
*pathname*/functions.import.php on line 102"

Is this an ImageMagick or Tesseract function that I should be looking for,
or is it an internal function to QueXF? Is it even a function?

Thanks in advance. Looking forward to increasing the processing power of
this already great piece of software.

-Nate

Re: [Quexf-discuss] Example output file?

[Adam Z. <az...@de...>]

Dear Brian,

The output of queXF is a fixed width ASCII file and a corresponding DDI
v2 (see: http://ddialliance.org) file.

I have attached the output of the 2 test forms in the "filled" directory
of the doc directory of the queXF distribution.

The fixed width ASCII filed produced is attached as test.dat and the DDI
file is test.xml

Regards,
Adam Zammit

Brian Binovsky wrote:
> I'm looking for something to markup the output of a fairly large database.
> I want to use the QueXF OMR capability to essentially survey the
> correctness of the database entries.
>
> However, I was hoping someone could save me some headache and show me
> what the output of QueXF looks like...for example for the test
> questionaire?
> I want to see the output so I can see it's suitability for me to use
> to correct the original data source.
>
> I found the example test in the .zip provided, but seemingly not a
> sample result file.
>
> Thanks
> BB
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> Quexf-discuss mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quexf-discuss
>
>   

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9251 7632 International: +61 3 9251 7632
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

[Quexf-discuss] Example output file?

[Brian B. <tec...@gm...>]

I'm looking for something to markup the output of a fairly large database.
I want to use the QueXF OMR capability to essentially survey the
correctness of the database entries.

However, I was hoping someone could save me some headache and show me
what the output of QueXF looks like...for example for the test
questionaire?
I want to see the output so I can see it's suitability for me to use
to correct the original data source.

I found the example test in the .zip provided, but seemingly not a
sample result file.

Thanks
BB

Re: [Quexf-discuss] Looking for some help with QueXF

[Adam Z. <az...@de...>]

Dear Nate,

It does sound like the problem is with the re-scaling from 8.5x11 to A4
within queXF.

If you can send me the following information, I will try to fix the
issue from this end:

*PDF of your survey that you generated from PDF tool on our tools webpage
*PDF of your scanned in survey (2 or 3 different filled in ones if you
have them would be best)

Also, please make all future requests for assistance to the queXF
mailing list (see: http://quexf.sourceforge.net/mailinglist.php ) as
others may assist you or be assisted by your solution.

Regards,
Adam Zammit

Nate Straight wrote:
> Hello again,
>
>  I managed to get past my Ghostscript problems, but now I have another
> problem.
>
> I can't figure out how to use my own printed and scanned-in surveys,
> because I keep running into an error trying to read the scanned
> surveys. I am using the QueXML to PDF tool on your tools webpage to
> convert my QueXML files to printable surveys. I think the problem is
> that the online tool makes a survey PDF in A4 size (rather than my
> local 8.5 x 11" size). I can import and band the survey fine
> (actually, I'm also having a serious problem with horizontal banding,
> but that's something I can work around I guess), but when I try to
> upload my completed surveys, it gives me a "cannot find QID" error
> message. I have a feeling the problem is with the resizing of the
> survey page that necessarily happens when I print the A4 survey to fit
> on 8.5x11 paper and then scan it back in. Since things get rescaled
> and resized in the printing, the question boxes and barcodes aren't
> where the banded survey expects them to be. Does this sound like the
> right analysis of the problem? Is there an easy way to get around this?
>
> Thanks,
>
> Nate Straight
>
> On Fri, Sep 19, 2008 at 12:04 PM, Nate Straight
> <nat...@gm... <mailto:nat...@gm...>> wrote:
>
>     Hi there,
>
>     I'm the assessment coordinator for the college of business of a
>     private university. I'm looking to use QueXF (or something like
>     it) to handle some surveying / data entry tasks to make our
>     data-gathering processes smoother. I'm having a lot of trouble
>     installing / using the QueXF package. I can get to the XAMPP page,
>     setup my database with the PHP admin, and access the QueXF admin
>     control panel, but when I run the "testconfig.php" script it says
>     that "gswin32c.exe exists but can not execute it. Please make sure
>     you are pointing to the executable file, not just the directory of
>     Ghostscript." I have ensured that I am pointing to the right
>     executable. I have double-checked that the executable functions
>     properly by running it on its own in a command shell. I have set
>     the windows' environmental variable "PATH" to include the binary.
>     Nothing works; it still won't load. Any ideas? Thanks in advance.
>
>     -Nate Straight
>
>

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9244 6370 International: +61 3 9244 6370
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

[Quexf-discuss] Announcing the release of queXS - An Open Source, Web Based CATI System

[Adam Z. <az...@de...>]

queXS - an Open Source, web based CATI (Computer Assisted Telephone
Interviewing) system, version 0.9.0 (Beta) has been released today.

See the website for more details:

http://quexs.sourceforge.net/

I would greatly appreciate if anyone is interested in Beta testing. I
will be more than willing to help with installation and bug squashing.

Please make requests for assistance with queXS to the queXS mailing list:

que...@li...

Regards,
Adam Zammit

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9244 6370 International: +61 3 9244 6370
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

Re: [Quexf-discuss] Importing forms and banding

[Adam Z. <az...@de...>]

Hello Martin,

1.

The view formboxesgroupsbelow90 is not be required any more so this 
warning can be safely ignored. The quexf.sql file creates real tables of 
each view and then drops them and over writes them with a view at the 
end of the statement. It does this to avoid any dependency problems. I 
will remove it in the next bug fix version.

2.

The constants in functions.image.php are set to handle a 300DPI 
monochrome (1 bit per pixel) file exported by Ghostscript.

Range checking should not be required, as each function that calls 
imagecolorat() is passed an area to read on the image that should exist.

I think the issue here may be that the system assumes that the scanned 
image will be a standard A4 page, and therefore at 300DPI will be 
2480x3508 pixels

Instead of changing the constants/ranges in functions.image.php - please 
try the following modifications:

functions/functions.import.php approx line 338:

exec(GS_BIN . " -sDEVICE=pngmono -r300 -sOutputFile=$tmp%d.png -dNOPAUSE 
-dBATCH $filename");

to:

exec(GS_BIN . " -sDEVICE=pngmono -r300 -sPAPERSIZE=a4 -dPDFFitPage 
-sOutputFile=$tmp%d.png -dNOPAUSE -dBATCH $filename");


admin/new.php approx line 46:

exec(GS_BIN . " -sDEVICE=pngmono -r300 -sOutputFile=$tmp%d.png -dNOPAUSE 
-dBATCH $filename");

to:

exec(GS_BIN . " -sDEVICE=pngmono -r300 -sPAPERSIZE=a4 -dPDFFitPage 
-sOutputFile=$tmp%d.png -dNOPAUSE -dBATCH $filename");


If you could also send through the PDF file you are using directly to me 
that would be great.

The recommended scanner setting is for Monochrome (1bit), 300DPI, A4.


3.

I think the issues you have with banding are due to the imported 
size/type of images - try again once you have made the changes above.

PNG's are dumped directly into a blob in the database, which can then be 
read using php's imagecreatefromstring() function - there shouldn't be a 
need to use adodb's functions.


I hope we can get to the bottom of these problems.

Regards,
Adam Zammit


Martin Smith wrote:
> Hello folks (I'm back :),
>
> I'm working on building a form and getting it working with queXF, and
> I've encountered a few troubles; I'm hoping perhaps someone has some
> suggestions or comments on how they use queXF (i.e. what works).
>
> Here's my list:
>
>  1. The data script database/quexf.sql has a warning with MySQL 5.0.54,
> specifically about "Unknown table 'formboxesgroupsbelow90'" when it
> tries to drop a view, if it exists, before creating it. Seemed weird,
> but overall didn't cause any problems as I always ran the script on an
> empty database.
>  
>  2. functions/functions.image.php -- there's all kinds of constants in
> this file. I'm not sure they work with my images, and even worse, there
> is some missing range checking, so they don't just fail to find lines,
> they actually spit out errors. This biggest offender was
> "imagecolorat($image, $x, $y);" which I had to replace with
> "imragecolorat($image, min($x,imagesx($image)-1),
> min($y,imagesy($image)-1));". Once I checked ranges, it failed to find
> registration marks for cropping, but at least it didn't entirely blow up
> :).
>
>  3. admin/band.php:233 -- when I try to band a scanned PDF, I get all
> kinds of failures about not being able to pull the PNG data out of the
> database, e.g. "gd-png: fatal libpng error: Read Error: truncated data",
> "gd-png error: setjmp returns error condition", "Passed data is not in
> 'PNG' format". But showpage.php has no trouble with it -- I only get
> errors once I've selected a field box. Am I scanning at the wrong
> resolution or otherwise producing a file that GD can't compose with the
> red field indicator box? If I use the cleanly generated empty/blank PDF,
> it works great. When I use a higher quality scan, it can't find the
> barcode. When I use the lower quality scan, it breaks on the PDF images
> when banding (when it tried to alpha overlay the red boxes, I think).
>
> Overall, the storage mechanism for the PNGs looks like it should be
> using ADOdb's updateBlob(), and maybe that is causing some of my
> problems, though I can't figure out why cleanly generated Apache FOP
> PDFs work, and my scanned one blows up *only* once I select a field (...
> is it possible the image inside the scanned PDF isn't PNG, and the
> conversion to PNG fails? or I convert at a different dpi? or gd fails
> the overlay?).
>
> Let me know if I can send anyone the PDF I'm using, or if there's a
> recommended scanner setting (dpi, size, etc) for this kind of thing.
>
> Thanks,
>
> Martin Smith, Systems Developer
> ma...@be...
> Bureau of Economic and Business Research
> University of Florida
> (352) 392-0171 Ext. 221 
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
> Don't miss this year's exciting event. There's still time to save $100. 
> Use priority code J8TL2D2. 
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
> _______________________________________________
> Quexf-discuss mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quexf-discuss
>
>   

-- 
Adam Zammit, Research Fellow in Survey Technology

Deakin University 221 Burwood Hwy, Burwood Australia.
Phone: 03 9251 7290 International: +61 3 9251 7290
Fax: 03 9244 6370 International: +61 3 9244 6370
Email: ada...@de...
Website: http://www.deakin.edu.au/buslaw/dcarf/
Deakin University CRICOS Provider Code 00113B (Vic)

Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone.

Deakin University does not warrant that this email and any attachments are error or virus free.

[Quexf-discuss] Importing forms and banding

[Martin S. <ma...@be...>]

Hello folks (I'm back :),

I'm working on building a form and getting it working with queXF, and
I've encountered a few troubles; I'm hoping perhaps someone has some
suggestions or comments on how they use queXF (i.e. what works).

Here's my list:

 1. The data script database/quexf.sql has a warning with MySQL 5.0.54,
specifically about "Unknown table 'formboxesgroupsbelow90'" when it
tries to drop a view, if it exists, before creating it. Seemed weird,
but overall didn't cause any problems as I always ran the script on an
empty database.
 
 2. functions/functions.image.php -- there's all kinds of constants in
this file. I'm not sure they work with my images, and even worse, there
is some missing range checking, so they don't just fail to find lines,
they actually spit out errors. This biggest offender was
"imagecolorat($image, $x, $y);" which I had to replace with
"imragecolorat($image, min($x,imagesx($image)-1),
min($y,imagesy($image)-1));". Once I checked ranges, it failed to find
registration marks for cropping, but at least it didn't entirely blow up
:).

 3. admin/band.php:233 -- when I try to band a scanned PDF, I get all
kinds of failures about not being able to pull the PNG data out of the
database, e.g. "gd-png: fatal libpng error: Read Error: truncated data",
"gd-png error: setjmp returns error condition", "Passed data is not in
'PNG' format". But showpage.php has no trouble with it -- I only get
errors once I've selected a field box. Am I scanning at the wrong
resolution or otherwise producing a file that GD can't compose with the
red field indicator box? If I use the cleanly generated empty/blank PDF,
it works great. When I use a higher quality scan, it can't find the
barcode. When I use the lower quality scan, it breaks on the PDF images
when banding (when it tried to alpha overlay the red boxes, I think).

Overall, the storage mechanism for the PNGs looks like it should be
using ADOdb's updateBlob(), and maybe that is causing some of my
problems, though I can't figure out why cleanly generated Apache FOP
PDFs work, and my scanned one blows up *only* once I select a field (...
is it possible the image inside the scanned PDF isn't PNG, and the
conversion to PNG fails? or I convert at a different dpi? or gd fails
the overlay?).

Let me know if I can send anyone the PDF I'm using, or if there's a
recommended scanner setting (dpi, size, etc) for this kind of thing.

Thanks,

Martin Smith, Systems Developer
ma...@be...
Bureau of Economic and Business Research
University of Florida
(352) 392-0171 Ext. 221