GeekTips

original is on bottom and image with text is very light and hard to read. ScanTailor with Otsu and other algorithms darkens the text as seen on top image.

49 views20:41

GeekTips

Use xreader instead of evince (no zooming thumbnails) for PDF reader to zoom in on thumbnails to quickly locate chapters to assign page numbers to them for booky.sh

In this case I typed them out and didn't both capitalizing anything which saved me some time.

56 views00:04

GeekTips

convert to title case with article exceptions and notice : after chapter number
titlecase -f aaa_index.txt -o titlecase.txt

53 viewsedited 00:06

GeekTips

generate index for the PDF with booky.sh
booky.sh output_ocr.pdf titlecase.txt

57 views00:08

GeekTips

This bash script generates bookmarks automatically as shown on the screenshot on the right. On the left PDFSam Basic let's you choose whether to retain existing bookmarks or not. Here the Bible already had bookmarks so it retains them.

Quite impressive script to quickly get bookmarks. I'll spend a little time trying to figure out how to retain bookmarks but most likely that's over my head.

combine multiple PDFs into a single PDF
Create one bookmark (filename) for each PDF in directory.
Won't retain existing bookmarks like PDFSam
author Mateen Ulhaq

62 views02:58

GeekTips

#!/bin/bash

out_file="combined.pdf"
tmp_dir="/tmp/pdftk_unite"
bookmarks_file="$tmp_dir/bookmarks.txt"
bookmarks_fmt="BookmarkBegin
BookmarkTitle: %s
BookmarkLevel: 1
BookmarkPageNumber: 1
"

rm -rf "$tmp_dir"
mkdir -p "$tmp_dir"

for f in *.pdf; do
    echo "Bookmarking $f..."
    title="${f%.*}"
    printf "$bookmarks_fmt" "$title" > "$bookmarks_file"
    pdftk "$f" update_info "$bookmarks_file" output "$tmp_dir/$f"
done

pdftk "$tmp_dir"/*.pdf cat output "$out_file"

62 viewsedited 02:58

GeekTips

This is pretty cool.... Adobe renamed ClearScan in 2015 to Edit Text in Image. It basically creates a anti-aliases around the text and vectorizes the text. Many hours spent try to get this work and will spare you all the details of tried this and that but this is what did actually work:

Download about 700MB and installs 1.8GB of texlive. This is a LaTex package as recommended by pdfsak. You may skip the fonts and extra and see if it works but get base and potrace are for sure required.

sudo apt install texlive-latex-base texlive-fonts-recommended texlive-fonts-extra texlive-latex-extra

potrace

download appimage of magick (ImageMagick)
right click and Permissions | check to Execute allow

sudo cp magick /usr/bin/

Now install pdfsak in python

pip3 install —upgrade pdfsak

Acrobatusers

Better PDF OCR: ClearScan is smaller, looks better

In this tutorial, learn about the advantages and disadvantages of ClearScan over Searchable Image OCR in Acrobat 9.

61 views13:39

GeekTips

pdfsak -if input.pdf -o clearscan.pdf --clearscan

screenshots:

original on top (blurry)
clearscan on bottom

A few years ago I used Adobe Acrobat DC Pro back when I was on a mac and remember it could replace the image with text by creating a custom font on-the-fly and create a much much smaller file size. Still need to manually go through and check the accuracy of the words so it's still somewhat time consuming. This doesn't do that but still it really helps clean up old PDFs.

55 viewsedited 13:39

GeekTips

Screenshot is showing zoomed in after cleanscan.  Notice many tiny dots.

Best to despeckle it (remove the tiny dots) in ScanTailor Advanced (unpaper is over my head). So some PDFs can do clearscan first then feed to ScanTailor but old ones with crappy background and double pages, etc. will have to ScanTailor it. img2pdf it to a pdf then clearscan them back to ScanTailor to despeckle then to ocrmypdf.

57 viewsedited 13:39

GeekTips

despeckle of 2.5 in ScanTailor when used with clearscan looks like it's removing ALL dots. 2.0 almost removed all dots (marked in red) say 95% but not all. Default is 1 and max is 3.

48 views13:39

GeekTips

This one is the true test.

screenshot
top one is ScanTailor only

bottom one is clearscan (not cleanscan..get confused with ocrmypdf which has a --clean option) and then processed with ScanTailor. File size is a tad bigger but I believe the improvement is worth it.

48 viewsedited 13:39

GeekTips

ScanTailorOnly.pdf

109.5 KB

ScanTailorOnly.pdf is 109KiB
ClearScan-ScanTailor.pdf is 272KiB

need to do these on an entire book to see if it's double or triple in size. If so guess it may not be worth it.
edit: ClearScan not CleanScan
edit2: seems like I won't use ClearScan only when the image is rather good to start with. Meaning the text in the image isn't blurred all that much.

CleanScan-ScanTailor.pdf

272.9 KB

47 views13:39

GeekTips

This PDF is already OCR'd but many times it isn't or perhaps you just wish to OCR an image of anti-Termites. But multi-column text is tricky to copy both in PDF and image OCR.

46 viewsedited 14:20

GeekTips

The cool thing here is the PDF annotation, note-taking app Xournalpp the nightly build version now has Select text either Linear or in a Rectangle. I just downloaded the appimage and it works great.

50 views14:20

GeekTips

NORMCAP and TextSnatcher both OCR images. Looks like NORMCAP does a tad better especially with line breaks. Both are good to have at your disposal though.

54 views14:20

GeekTips

PDF with scanned images that have varying image resolutions. This won't work and create chaos, havoc in ScanTailor with Margins and Select Content even.

47 views23:27

GeekTips

PDF with non-uniform images. Resize larger images smaller ones maintaining aspect ratio to solve this problem.

Smaller ones are: just a few examples
1501 x 1152
1491 x 1155
1547 x 1159
1514 x 1159

Larger ones are: just a few examples
4649 x 3610
4641 x 3647
4677 x 3597
4849 x 3819

Problem with this is in ScanTailor Advanced the smaller images remain small compared to the larger resolution images after setting margins. This pdf is landscape with two pages scanned per page.

A4 = 210mm x 297mm
A4 + A4 = A2 = 420mm x 594mm

I thought just using this scale-to option would solve the trick but it didn't.

pdftoppm -r 300 -scale-to 842 -tiff -tiffcompression deflate nonuniformimages.pdf dump/img

so instead just did this and didn't specify dpi

pdftoppm -tiff -tiffcompression deflate nonuniformimages.pdf dump/img

44 viewsedited 23:27

GeekTips

sort by size to select the larger resolution images and move them to a new directory.

mogrify will modify the existing images so make a quick copy of them just in case. ImageMagick doesn't have deflate compression so have to use LZW.

-resize 555 resizes width and -resize x555 does height.

mogrify -resize x1172 -compress LZW dump3/*.tif

They maintained their aspect ratio and the new resolution on a few of them are:
1510 x 1172
1519 x 1172
1516 x 1172
copy these back into your /dump directory and ScanTailor will process just fine now.

40 views23:27

GeekTips

You can try to align chapter numbers by adding blank pages or deleting some unwanted pages so the table of content numbers match the actual PDF page number.

Or you can use vim and a plugin called vim-NumUtils.

sudo apt install vim

download vim-NumUtils and after unzipping put the doc/ and plugin/ into the ~/.vim directory. It can do

NumUtilsAdd, NumUtilsSub, NumUtilsMul, NumUtilsDiv

Here all the actual PDF page numbers are 3 pages behind the listed page numbers.

:% NumUtilsSub 3, '\,'

sometimes I type a space after the , so to get those too do

:% NumUtilsSub 3, '\, '

Quick vim commands
:wq - write/save file and quit
:q - quit
u - undo
Ctrl-R - redo
i - insert mode ESC to quit insert mode
a - append text
% is a range for entire document
Great vim guide

41 viewsedited 23:27

GeekTips

{
foreword by the author,6
{
poem:,
first, my country, 15
this is my own, my dear, and native land,16
god made me free, 21
to the stars and stripes,33

Renumber chapter page numbers in text file with math operations. How to do it with sed and

awk

sometimes I'll type a space after the , so let's substitute space after a comma followed by a number then the 2nd substitution separated with a ; does same but has no space

sed -i -r 's/(, )([0-9])/,\2/; s/,([0-9])/%\1/' toc.txt

-i insert in place the /s (substitutions) looking for , followed by any number of digits. Some chapter titles have , in them and it'll ignore them
-r extended regex so don't need to escape capturing groups
\2 2nd capture group put a , then single digit to remove space
\1 is the first capture group ([0-9]) which is just a single digit and re-inserting it after the %
without touching other commas in chapter names

{
foreword by the author%6
{
poem:,
first, my country%15
this is my own, my dear, and native land%16
god made me free%21
to the stars and stripes%33

Now subtract 3 from the chapter page numbers

awk -F% '{if (/%/) {print $1 "," $2-3} else {print $0}}' > renumbered.txt

-F% field separator , would work but if the chapters contain , which they do sometimes it screws up fields so needs to be something else like

%
if (/%/)

there's a match for % on the line then
print $1 print field 1 which is the text string up till the

%
","

will print a comma between $1 and $2 so it's ready for

titlecase
print $2-3

print the number after the % and subtract 3 can do + addition * multiplication / division also
else print $0 prints entire line if no match

changes to:

{
foreword by the author,3
{
poem:,
first, my country,12
this is my own, my dear, and native land,13
god made me free,18
to the stars and stripes,30

now it's ready for title case

titlecase -f renumbered.txt -o chapters.txt

changes to:

{
Foreword by the Author,3
{
Poem:,
First, My Country,12
This Is My Own, My Dear, and Native Land,13
God Made Me Free,18
To the Stars and Stripes,30

combine all three commands on one line

sed -i -r 's/(, )([0-9])/,\2/; s/,([0-9])/%\1/' toc.txt && awk -F% '{if (/%/) {print $1 "," $2-3} else {print $0}}' toc.txt > renumbered.txt &&  titlecase -f renumbered.txt -o chapters.txt

then booky to add chapters to pdf

booky.sh SomeBook.pdf chapters.txt

If you want to offset the pages numbers by +5 just change $2-3 to $2+5

sed -i -r 's/(, )([0-9])/,\2/; s/,([0-9])/%\1/' toc.txt && awk -F% '{if (/%/) {print $1 "," $2+5} else {print $0}}' toc.txt > renumbered.txt &&  titlecase -f renumbered.txt -o chapters.txt

changes to:

{
Foreword by the Author,11
{
Poem:,
First, My Country,20
This Is My Own, My Dear, and Native Land,21
God Made Me Free,26
To the Stars and Stripes,38

44 viewsedited 04:06

About

Blog

Apps

Platform