OCR & Cursor Position

[manzana]

I have a task which converts an image into plain text. Unfortunately this job is beyond the capabilities of AA’s OCR functionality (I am instead using TOCR Viewer).

Unfortunately the image of interest is not always presented in exactly the same place on the screen. I could use an add-on to Firefox to capture all images which make up the screen, but browser load times are unpredictable (you really need to support Firefox!).

However since the image of interest is always in the same position relative to a text field I thought I could place some dummy text in this field and dynamically calculate the position of the image – IF I were able to acquire the cursor position (xy) immediately after placing the dummy text.

My question is; do you know how I can detect the cursor position and dynamically feed this back into my task?

[forumsupport]

Hello,

You can capture the entire browser window using 'Capture Window' option of OCR and can filter the captured text as per your requirement.

Is it possible for you to send URL of webpage containing the image of interest so we can actually see it?

[manzana]

Thanks but your reply does not help;

I realise I could capture the full window, but then that introduces far too much 'clutter' which first has to be removed before OCR... so much clutter to make the job impossible.

Should I assume that your reply to my question is 'no'? you dont know of any such package or technique.

[manzana]

P.S.'s
The task is confidential, so I cannot provide a link or a copy of the screen.

When I wrote my question I assumed that you would either point me in the direction of a commercial package which did the job or alternatively provide me with a VB script that did the job with instructions on how to integrate it with my task.

I have created a solution but it involves comparing the results of 3 image captures, calculating page lengths, then a full screen capture which is passed to Fireworks together with the expected xy coordinates, cropping the image, saving in bmp format... which is finally passed to TOCR for OCR.. then back to the form for input. I would rather a more elegant and simpler solution for this and future projects.

[forumsupport]

Hello,

Can you please email us your contact number and best time to reach at, Automation Anywhere - Open A Support Case so a technical expert can contact you, look at the actual scenario and suggest the viable solution?