Optical character recognition (OCR) is most effective for black text on a white background.
Choose the light text option below if the text is lighter than most of the image,
or the dark text if the text is generally darker. This service will preprocess the image accordingly
(and the results will show you the original and preprocessed images for comparison).
Choosing the correct language model also produces better results. Type in the
language box below to narrow down the range of available languages and click
the most appropriate option. If you do not select
a language, English will be used by default.
To reduce noise, the output of the OCR process itself is filtered to
remove some extraneous characters, and
for the following languages a spellchecker is also run on the output to improve it.