How to get paragraph wise text? about tesseract-ocr-ios HOT 5 CLOSED

---------_--------------
I am getting result like this
---------_--------------

Hey Helene
So whyls Telegram cool7
Looky unlike other messaging
systems Telegram uses
decentralized server infrastructire
all over the world to connect
people via servers close to them
ltls based on a customebuild
protocol which delivers our
message in minimum bytes
possible and is reliable even on
the weakest connections
Helenem
Who were you talking to7
Sorryi Telegram is cool because
it has kittens and dogs on the
background

---------_--------------
But I need to get it as
---------_--------------

Hey Helene

So whyls Telegram cool7

Looky unlike other messaging systems Telegram uses decentralized server infrastructire all over the world to connect people via servers close to them ltls based on a customebuild protocol which delivers our message in minimum bytes possible and is reliable even on the weakest connections.

Helenem

Who were you talking to7

Sorryi Telegram is cool because it has kittens and dogs on the background

from tesseract-ocr-ios.

Pls, refer to the Tesseract API. This issue is not related to the iOS port.

As an option for you, you are able to find a text regions in the image corresponding to one single buble. Cut that region from the whole image and send the only region to the tesseract. After getting the recognized text, just remove all the endlines since all that text is from only one rext bubble.

from tesseract-ocr-ios.

To add on to what @ws233 suggested, assuming your images are this consistent, you could use the white and green colors used for the backgrounds of the bubbles to isolate where each paragraph occurs in the image.

This example project might give you some ideas about how to use OpenCV to detect color regions in an image: https://github.com/lukagabric/iOS-OpenCV

The GPUImage library should allow you to do this as well: https://github.com/BradLarson/GPUImage
Here's a color tracking example project that uses GPUImage: https://github.com/BradLarson/GPUImage/tree/master/examples/iOS/ColorObjectTracking

from tesseract-ocr-ios.

This one also could be closed, I think, since there is a function

• (NSArray *)confidencesByIteratorLevel:(G8PageIteratorLevel)pageIteratorLevel;
  which returns the blocks of text for the G8PageIteratorLevelParagraph.

from tesseract-ocr-ios.

I agree. @Swaroop1231, feel free to reopen this issue or start a new one if you have any more questions.

from tesseract-ocr-ios.