FAQ

Versions

Image Processing

Export

Optical Character Recognition (OCR)

  • We checked the accuracy of the OCR engine in the 10 available languages. Even with pictures taken in poor circumstances (document on several other documents, picture taken in front of the document rather than above...), we got very good results.

    Here are some recommendations for improved OCR results:
    • you should be right above the document you want to photograph.
    • use the camera flash when you take a picture (iPhone 4 only).
    • make sure the text to be recognized is parallel to the horizontal blue lines of the on-screen grid.
    • choose the language of the document in the list (you can recognize a letter in Spanish while using Prizmo in English).
    • optimize the character contrast. Zoom in to get the text area in close-up, and play with the optimization slider. You should be able to read the text areas on the whole page.


    Here are some additional recommendations for taking pictures:


    The text is perspective distorted. As it is not perfectly straight, it will lower OCR accuracy.

    The text is tilted. As it is not perfectly straight, it will lower OCR accuracy.

    Use the perspective tool to make the picture straight before OCR processing.










    The image is blurry.
    As a result, OCR will perform badly.

    Contrast is poor.
    You'll get bad OCR results.

    For proper results, use the voice command to avoid motion while shooting. Use the flash of your iPhone 4 if lighting is bad.










    If your picture is blurry,
    you'll get something like this.

    A photograph without perspective correction will lead to a poor OCR performance.

    With a good shot,
    you'll get very accurate OCR results.





Text-to-Speech (TTS)