jasenmichael / clean-extract-text

clean image(or pdf) text, then use tesseract ocr to extract text from cleaned image.

clean-extract-text

a simple starter for ocr from png/jpg/pdf files - "with receipt text extraction" in mind

clean image(or pdf) text, then use tesseract ocr to extract text from cleaned image.

git clone https://github.com/jasenmichael/clean-extract-text.git
cd clean-extract-text
npm i
npm run start

to clean and extract text from your own png, jpg, or pdf edit the main.js

// edit file 
let file = 'src/receipt.pdf'
// and replace 
let file = '<your-file>.pdf|png|jpg'

requires imagemagick

# on debian linux
sudo apt-get install imagemagick

# on iOS
brew install imagemagick

READING PDF files

to convert .pdf files with imagemagick, you must edit the policy.xml file, it is probably located at - /etc/ImageMagick-6/policy.xml

replace

  <policy domain="coder" rights="none" pattern="PDF" />

with

  <policy domain="coder" rights="read|write" pattern="PDF" />