Also the magenta colour on black could be problematic - try outputting the
intermediate Tesseract image to see how this gets turned into the input for
Tesseract (post just last week on how to do this) to understand if you
should pre-process your image e.g. if this is always the position of the
pati
The font is very unusual - pixellated edges, close together and far too
small. You might want to also try a different page segmentation parameter
since the image contains a big blob of non-text.
On 28 June 2016 at 10:44, wrote:
> Hello,
>
> tesseract-ocr is not identifying the words from the att
2 matches
Mail list logo