[Staffslug] OCR Script for PDF Pages

Hi: Using Debian, I found I had several OCR Packages. They're probably on DVD #2 still. I tried them on a big PDF file with illustrations. "Tesseract" was best, but the result would still need manual editing with a Spelling Checker. "gocr" and "ocrad" were available too. Note that with all these you only get the text. The illustrations were nowhere to be seen. And looking again, I remembered I'd written a little bash script for OCR. The reason for this is that you probably need to burst the PDF into separate pages using "pdftk", then covert them into another format (using "convert") to get the type of input file required by your OCR Package. ====================================== #!/bin/bash for page in *.pdf do convert -density 300x300 $page ${page%pdf}pbm ocrad -o ${page%pdf}txt ${page%pdf}pbm rm ./*.pbm done ====================================== My script got edited several times, ending up with the "ocrad" version as you see, using PBM files as input. You need to run this within a directory full of PDF pages, numbered with leading zeros so the processing runs in the correct order. If you try to process a big PDF as one unit, you may get nowhere, whereas with separate pages you can see how your OCR is progressing. Yours truly: Frank Mitchell On Friday 06 Dec 2013 14:25:26 Jonothon Nihill wrote: