Signature, signature, who can grab the signature?
I'm looking to identify "if a signature has been included in a document". I am not concerned at grabbing the actual verbiage, just a simple Boolean Y or N.
I don't see any entries on Grooper X Change discussing this.
I would assume it is using a blank line as a baseline, with a specific dpt count, and "if dpi count surpasses baseline, then "Y", otherwise "N". Example below
Best Answer
-
OptionsRandoCalrisian Posts: 195 adminWith the example provided in this thread, I'll show how I set this up to have a Data Field return a True value.
I first created a Batch and copy/pasted the image into it, which created a page object. I also added the page object to a folder.
Next I used the Full Text - Blended OCR Profile to get character data.
I then created a Content Model and added a Data Model and a Document Type.
I need to generate a Form Type for the Document Type, as this is required to establish OMR Zones and Registration Zones on a Data Element Profile. To do this, I need to "train" a version of the Document Type. To gain access to this functionality, I set the Classification Method of the Content Model to Lexical. From there, an extractor has to be established so I set the Text Feature Extractor • Type to Internal, and I wrote a very simple pattern:\w+
This pattern could have been anything at all, as I'm not needing actual Weightings or anything, just access to create the Form Type.
Having set the previous properties, I could then go to the Classification Testing tab, make sure I'm pointed at the correct Batch, select the Folder/Document, and use the Train Document command.
This creates a Form Type, which is a child object of a Document Type. Doing this will store images (and weightings of the trained features ... which we don't care about in this specific case) of all the Document's Pages, within this Form Type object.
Next I created a Data Field, set its Value Type to Boolean, and Minimum Confidence to 0%.
The next bit of functionality comes from a Data Element Profile, which is essentially a set of overrides applied to a Document Type.
With the Document Type selected, I went to the Data Element Profiles tab, selected the sig_test Data Field, pressed the green plus button (which adds a Data Element Profile to the Data Field, and also adds a little green dot to the Data Field {instead of the default gray one} to signify that this Data Field now has a Data Element Profile), and set the Override Extractor property to True.
Next I used the small Add drop-down next to the +/- buttons, and within that menu selected the OMR Zone option. I then used the selection tool on the Page Viewer to drag out the corners of the green OMR Zone to the area I want. OMR means Optical Mark Recognition. This is a zonal technique that is a type of legacy approach to reading check boxes and radial buttons on documents. It essentially looks within an arbitrary zone (or box) to understand a fill threshold of pixels. In this case, the default OMR Threshold of 10% should work fine.
You can use the Test the selected zone. button to bring up the OMR Zone Unit Test Results window. In this tester you can see that the percentage fill of black pixels is 12%, which is above the 10% OMR Threshold, so it returns a result of True.
In order for this to work properly, it can't just be a zone on a page. It needs to be anchored off some nearby text. To do this I used the Add drop-down again (with the OMR Zone selected), and chose Registration Zone. This adds a yellowish box, which you want to open up to encapsulate some text near the area you're wanting the OMR Zone to be used. From there I set the Extraction Type property to Full Text, and set a Registration Pattern of:
Signed for identification
I then clicked the Test the selected zone. button for this Registration Zone to bring up the Registration Zone Unit Test Results window. I can see this pattern and zone works, as there's a blue box with a red dot signifying the value and it's position on the document. From there I clicked the Save Location button to save this result.
With this Data Element Profile established, I was able to then go back to the Data Model, Test Extraction, and see that my Data Field returns a result of True.
Randall Kinard
rkinard@bisok.com5
Answers
Let me set that up now, with this example...
rkinard@bisok.com