OCR snipping tool with imagemagick and tesseract

inspired by normcap which i like very much but seems slow to launch on my pc. this uses slop to select a region of the screen, imagemagick to capture and process it into a blackish-text-on-whiteish-background image for tesseract to convert to text, xclip to load it to the clipboard, and notify-send to preview the output.

there are two versions, one that saves an image, text file, and loads it to the clipboard, and another that only loads it to the clipboard.

#!/bin/bash

# setup

timestamp=$(date +%Y-%m-%d--%H-%M-%S)

slop=$(slop -f "%x %y %w %h %g %i") || exit 1

read -r X Y W H G ID <<< $slop

# take and process pics

import -display :0 -window root -crop "$W"x"$H"+"$X"+"$Y" $HOME/Pictures/screenshots/$timestamp.png

convert -colorspace gray $HOME/Pictures/screenshots/$timestamp.png $HOME/Pictures/screenshots/$timestamp-temp.png

# invert pic if dark so tesseract can read it

value=$(identify -format "%[mean]" $HOME/Pictures/screenshots/$timestamp-temp.png |rev| cut -d"." -f2- |rev)

if [[ $value -lt 19000 ]];

   then

      mogrify -channel RGB -negate $HOME/Pictures/screenshots/$timestamp-temp.png

fi

# ocr

tesseract -l eng $HOME/Pictures/screenshots/$timestamp-temp.png $HOME/Pictures/screenshots/$timestamp

cat $HOME/Pictures/screenshots/$timestamp.txt | tr '\n' ' ' | xclip -i -selection clipboard

rm $HOME/Pictures/screenshots/$timestamp-temp.png

notify-send "📋 OCR processed:" "$(xclip -selection clipboard -o | head -3)"

#!/bin/bash

# setup

timestamp=$(date +%Y-%m-%d--%H-%M-%S)

slop=$(slop -f "%x %y %w %h %g %i") || exit 1
read -r X Y W H G ID <<< $slop

# take and process pics

import -display :0 -window root -crop "$W"x"$H"+"$X"+"$Y" /tmp/$timestamp.png

mogrify -colorspace gray /tmp/$timestamp.png

# invert pic if dark so tesseract can read it

value=$(identify -format "%[mean]" /tmp/$timestamp.png |rev| cut -d"." -f2- |rev)

if [[ $value -lt 19000 ]];

   then

      mogrify -channel RGB -negate /tmp/$timestamp.png

fi

# ocr

tesseract -l eng /tmp/$timestamp.png stdout | tr '\n' ' ' | xclip -i -selection clipboard

notify-send "📋 OCR processed:" "$(xclip -selection clipboard -o | head -3)"

fruitcombo

OCR snipping tool with imagemagick and tesseract