Just find out there's bug in my image scale function. After fixed it the input image is surely scaled by 300% and tesseract works like a charm.
在 2016年11月2日星期三 UTC+8下午3:20:08，James Liu写道： > > Hey all, > > I'm in a task to OCR screenshot of terminal output. They are error logs > from standard ubuntu or centos systems. It's kind of easy compared with > images from real world camera. My result was so poor. I tried everything > found on google but nothing helps. Hope someone here could give me some > advise. Anything is welcome. > > My eng.traineddata was downloaded here > https://github.com/tesseract-ocr/tessdata . The image was preprocessed by > binarization and changing DPI 600*600. > > The command I run was : > tesseract -l eng test1_dpi_b.tiff out -psm 4 > > I also tested this image on a tesseract web service https://www.newocr.com. > Their results are much better than mine. > -- 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/43112351-0fe5-4a1a-8a4e-f4095d476d5a%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
