Hi Wickpick, What's your final goal? To collect still images from designated people occasionally visiting the meter installation site? Or placing e.g. an unattended camera to establish permanent meter monitoring? Depending on this approaches may vary.
For partially rotated numbers template matching is definitely a way to go. You may try to train Tesseract for partial digits - as an easier alternative - but I'm not sure it would perform well. Not quite clear - do you need to read the liter dials? If so then it isn't an OCR problem but rather a general computer vision problem. You would need to devise your own image processing pipeline along with the algorithmic approaches. I'm not sure it's worth implementing... Anyway, look what I've found as the first search result: https://etd.ohiolink.edu/rws_etd/document/get/case1258511722/inline A neat combo of simple math and algo techniques but accuracy in practice? Reflections are a thing that you certainly need to get rid of. Most of the time they render images unreadable even by humans. As for moist, it can present a problem as long as it's condensed vapor covering the entire or part of the glass. If it's mostly just liquid at the bottom then it's usually alright. And finally, I suppose Tesseract already has a pretty decent collection of trained fonts to work with most meter types. Regards, Dmitri Silaev www.CustomOCR.com On Tue, Jul 21, 2015 at 9:40 AM, Marc Bruins <[email protected]> wrote: > Hello all, > > I was wondering if it was possible to scan the cubical meters and the > liters on a water meter. > > There are a lot of circumstances possible where it is harder to scan the > image which are > > - The reflection of the camera on the glass > - Numbers can be half way down or up > - Most of the time these meters are found in very dark places > - It happens a lot that there is moist under the glass > - There are different fonts and sizes > > The meter which im talking about is this: > > <https://www.dwrcymru.com/eng/forms/meterreading/_images/meter.gif> > > So my question, is it even possible? If so what would you guys recommend > to create a good OCR for this (Which steps do i need to take?). > > Kind regards, > > Wickpick > > -- > You received this message because you are subscribed to the Google Groups > "tesseract-ocr" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/tesseract-ocr. > To view this discussion on the web visit > https://groups.google.com/d/msgid/tesseract-ocr/8c39934a-2c3c-4d3e-8355-466bf11eeec9%40googlegroups.com > <https://groups.google.com/d/msgid/tesseract-ocr/8c39934a-2c3c-4d3e-8355-466bf11eeec9%40googlegroups.com?utm_medium=email&utm_source=footer> > . > For more options, visit https://groups.google.com/d/optout. > -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/tesseract-ocr. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CAKzLxFOH7OeJfx_QHQrpgqbYpBc76c3ERcP%2Bg7eYxNyO8NVmWw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
