Browser-Zooming in to do Correct OCR with SeeTEst

1 reply [Last post]
ecoutin
Offline
Joined: 02/25/2012

Hi,

I have a problem. I need to validate some test a server returns on the Android browser it looks something like this:

"1234567"

But SeeTest is doing the OCR wrong, is confusing 6 with 0 because the text is TOO small in the browser. Here's how

The problem is that the initial conditions are that the browser is CLOSED(Not started on the device)

Once you load the server URL and send the query you are expecting a number like the one above and the server works fine BUT the browser loads and displays simply text, so the VISUAL level of the text is SO small that SeeTest thinks it's seeing 1234507 and FAILS the verification.

I made sure that's exactly the case using the debug function: Taking a scene of the browser with the 1234567 result and extracting the text result with OCR. Then manually Zooming in with the touchscreen and taking another scene and doing OCR again.

SeeTest returns 1234507 on the small text case and 1234567 on the Zoomed in Case.

How can I fix this?

Thank You.

support
Offline
Joined: 07/20/2011

Hi Einar,

there are three options for you in such case:

1. use the Color Filter in order to improve the OCR identification capability.

2. in application (unlike in your case) you can identify text by identify the Native id of the element and use the ElementGetText, and SeeTest will get the text from its Native properties.

3. the best option for you. when working on web, you can use the web commands. for your issue i would use the webGetText command:
Get the text of the element in focus (use the WebFind command to put an element in focus)

please notify me if you managed to work this out.