Read text from image data item

[keri_z_volans]

A citrix application we're using has a background gradient that interferes with font recognition, and makes OCR very unreliable. I am able to read the image, and edit it into black and white, so "recognise text" works on it. In order to use "recognise text" I am forced to launch a window displaying the image, attach to it, etc. Is there any way to use "recognise text" on a data item directly, or otherwise extract the text without launching a window?

[JiriPospisil]

Hi Keri, best way is probably surface automation - character matching. https://portal.blueprism.com/system/files/Introduction%20to%20Surface%2 You can map your own font and then read any value, it will take a bit more effort but I believe it is worth it.

[Denis__Dennehy]

Good answer by Jeri. Surface Automation is the only way to interact with Citrix - because Citrix has no technical/API adaptors like locally running applications. All that Citrix presents is effectively a picture/image of the applcation that is actually running on a Citrix server. It is therefore impossible to extract the data from Citrix 'without launching a window'

[keri_z_volans]

Thanks for the reply. I understand how Citrix, surface automation and defining fonts work. :) My question is whether it is possible to run ""recognise text"" on an image Data Item, rather than on a region in an active window.

[TomBlackburn1]

Keri, If you specify the font color, does that help Blue Prism to recognize the text from the background gradient. There is no method to recognize text directly on an image data item, only on a region. The only workaround I am aware of, you have already talked about in the OP. Depending on your limitations and objectives there are other various workarounds. Could you highlight the text and copy it to the clipboard? Do you need to recognize the text, or would an image comparison be sufficient? - Tom.

[Denis__Dennehy]

Hi Keri, I understand what you want to do now. It is possible using a prototype business object that basically calls the Blue Prism Character matching dll. You can pass it an image and the font details and it will do the Characther matching for you. As a prototype (it may eventually become a new internal business object) it is not guaranteed to work with future versions of Blue Prism. Please contact your Blue Prism mentor who will be able discuss exactly what you want to achive and why and give you this prototype object if needed..

[keri_z_volans]

That's perfect, thank you. Is there any documentation for the dlls?

[Denis__Dennehy]

There is no documentation, this prototype option is not Certified by the product team - so is available 'as is' for your experimentation.