Old fuzzy pages, still tricky at 1600 dpi with #XSane and #OCRmyPDF on unix.
=
Hooper Ranch Bookkeeper ............... cobhebgeneaneen Lisa Salkov, Mana Diaz (alt.)
=
1. Mana should have been Maria.
2. And a bunch of dots got halluncinated into random letters. Same as it ever was, back to encoded Bacon wrote Shakespeare gibberish.
Otherwise, damned decent!
@WorziArmin Ein Kollege hatte schon mal Tools fürs #Dokumentenmanagement vorgestellt. Aber ich fürchte: Das erfordert noch mehr Disziplin. #OCRmyPDF kann das Problem nicht lösen, das scannt ja nur ein und macht die Texterkennung. Für alle, die keine Lust haben zu sortieren, empfehle ich tatsächlich #Recoll. Festplatte indizieren, dann findet das fast alles. Aber mich würde das Chaos auf der Festplatte irre machen.
Ich bin ja sonst nicht so der Typ für #Software und Empfehlungen....
Aber das hier ist ein absolutes Muss, wenn Du massenhaft pdf-Dateien nachträglich mit einem Text-Layer versehen willst.
Massenhaft scannen in eine Datei und während der Texterkennung automatisch trennen lassen mit ist nur ein Highlight...
Muss man haben!
Github:
https://github.com/digidigital/OCRthyPDF-Essentials
#pdftk for terminal
@libreoffice draw
#masterpdf v4 is free; current version is paid
#ocrmypdf
#pdfunite etc
I can send you various aliases I have created. Also, see various pdf posts at linuxatty.wordpress.com.
Editing or redacting a #PDF using #LibreOffice Draw is far superior to the commonly used method of converting the PDF's pages into images and editing the images, because the latter results in a PDF that is many times larger and doesn't render as well. Also, text copy and paste is lost, which you can recover from to some extent with a tool like #OCRmyPDF, but you'll never get the text quality back to as high as it was before you converted the PDF to images.
#FOSS
Have you ever needed to extract text from images embedded in a #PDF? I can highly recommend the open source #CLI tool #OCRmyPDF which is easy to automate in for example a #DataPipeline.
It uses #Tesseract #OCR under the hood and has many options to experiment with to get the best possible accuracy for your language and PDF content.
You can get started with just a few commands:
https://samuelplumppu.se/blog/automated-text-extraction-from-pdf-images-with-ocrmypdf
Добавление OCR-слоя и другие преобразования PDF
При сканировании и сохранении в формате PDF зачастую документы сохраняются в виде графических изображений. Это неудобно, потому что делает невозможным полнотекстовый поиск по содержанию. Утилита OCRmyPDF решает эту проблему: она одной командой из консоли добавляет к PDF-документу слой OCR с распознанным текстом. Ниже упомянуты ещё несколько полезных инструментов для парсинга PDF, в том числе для преобразования сложных математических PDF-документов в текстовый формат Markdown.
2/2 re #OCR
All three were set to #rotate and #deskew. None rotated the page that was sideways, but they all #deskewed pages that needed it. Kofax was the speediest of the bunch, then #OCRmyPDF not far behind and #Foxit was by far the slowest.
File size Foxit produced the smallest file size, #Kofax created files double the original. OCRmyPDF struggled here, ballooning the original size by at least 6 times larger.
1/2 re OCR
I got to do a fun test at work with #OCR. #Foxit Phantom with AbbyFineReader from 2013, #Kofax Power PDF from 2020 and #OCRmyPDF via #WSL with #Tesseract as the OCR engine.
The best results in OCR were from OCRmyPDF great results. Second was Kofax lagging was the over 10-year-old Foxit. OCRmyPDF did perform great and just picked up a few more characters, especially fuzzy scanned text, plus it got some handwritten text.
Meine Fresse, sind wir heute wieder aktuell:
#PDF #OCR #CLI #Stapelverarbeitung ... als wäre es 2005 ;)
Wobei: Seit Dokumente zunehmend per Handy "gescannt" werden, könnte (nachträgliche) Texterkennung doch recht aktuell sein :)
https://www.tutonaut.de/pdf-texterkennung-stapelweise-fuer-windows-und-linux/
Цифровой архив с полнотекстовым поиском, в том числе по PDF и картинкам
У каждого человека с годами скапливается множество бумажных документов, в которых непросто разобраться или что-то найти. Эта проблема ещё более актуальна для организаций. Опенсорсная программа Paperless-ngx позиционируется как оптимальное решение для создания цифрового архива. Со встроенной системой распознавание символов (OCR) и обучением на основе ранее отсканированных документов она создаёт хранилище с поиском, где можно быстро найти любой документ. Всем документам присваиваются теги, так что они могут присутствовать в разных тематических категориях, это удобнее распределения по папкам. Paperless-ngx можно установить на домашний сервер и загружать документы через браузер с любого устройства.
https://habr.com/ru/companies/globalsign/articles/887176/
#Paperless #Paperlessngx #цифровое_хранилище #электронный_документооборот #сканирование_документов #OCRmyPDF
#WSL is nice because I can use #OCRmyPDF on #Ubuntu. I set it up to watch a folder for any new #PDF then automatically #deskew #rotate #OCR then #export to a "done" folder. It is very nice to have it done automatically in the background. No more opening, clicking to OCR and waiting on the software and unable to open other PDFs. Plus, this process is way lighter on resources. Man, I love #OpenSource.
#Tesseract
Lifehacker: This Free Tool Can Help You Search and Copy (Nearly) Any PDF. “There’s nothing worse than opening a PDF and realizing you can’t use the search function or even highlight text. This typically happens when a PDF was created by scanning a paper document — it’s just a series of images. Most modern scanning software uses Optical Character Recognition (OCR) so that words are both […]
For my server backend, I used a #python script to handle the requests.
Basically, it makes use of two components:
#CV2 (open computer vision). This is Swiss Army-knife for image manipulation. I use it to reduce an image to a b/w format which only contains the text. Ideal for a quick copy
#imagemagick to apply sigmoidal contrast, removing most problems due to poor illumination
#OCRmyPDF to make it more convenient to work with the scanned sheet
@nielso darf ich dir von unserem Herrn & Erlöser #PaperlessNGX (welcher auch die Wunder des #OCRmyPDF zum Nutzen seiner Jünger mehret) predigen? 😅
@nyx@im-in.space I just stumbled over this post and it got me thinking...
If you´re still on the lookout...perhaps #OCRmyPDF can help you.
https://github.com/ocrmypdf/OCRmyPDF
You would have to convert your documents to pdf and then throw them at OCRmyPDF to create searchable pdf files.
If you want a nice #selfhosted web ui, #PaperlessNGX uses OCRmyPDF internally. You can run it in a #Docker #Container, upload your pdf files into it and when it has done it´s thing, enjoy your searchable pdfs.
