It looks like you may need to fine tune train Tesseract on this particular font. From the letters in you images it looks like 'Bevan', which you can download from here:
https://www.fontsquirrel.com/fonts/bevan If you are unable to train Tesseract, I have sometimes had success by stretching (changing the aspect ratio) of the image. In this case it is quite a fat font so stretching it taller might improve the result. Hope this helps James On Tuesday, August 14, 2018 at 1:41:06 PM UTC+1, [email protected] wrote: > > It's interesting. I'v tried many way to process the img, binary inverse, > cut, resize. > I'v tried with oem of 3.0.0 and 4.0.0, psm of 3\6\7 > I thought maybe some one works, but actually no one did, and nothing went > out > Maybe this special fonts just hit some weakness of tesseract > > > 在 2018年8月14日星期二 UTC+8下午6:59:01，[email protected]写道： >> >> I use opencv to extract chars from image and combine them together, but >> tasseract failure to recognize it. >> I have tested with paramters "-c >> tessedit_char_whitelist=ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz,.-\\'" >> >> "-psm 7" and "-psm 8", still no lucky. >> please see attachment, ears.png >> >> but some others were successful, like godmother.png. >> >> who could teach me, please. >> > -- 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 https://groups.google.com/group/tesseract-ocr. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/9f8df077-3cd6-4e55-84ce-29dcb54f194a%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
