Someone used (probably) Adobe Professional OCR feature to OCR a scanned pdf file. The OCRed result had some errors, and it added the text result back to the text image in the pdf file, covering the actual text image, so I can't tell what the correct one that the text image shows
Can I recover the original image of text after processing by adobe professional OCR? Thanks.
You can easily list or extract all images from a PDF (or just from a specific page range) using the command line tool pdfimages. This tool is available for Linux, Unix, Mac OS X and Windows.
pdfimages -list -f 3 -l 7 my.pdf
Above command lists all the images from page 3 (-f "first") to page 7 (-l "last") without extracting them.
The most recent versions of pdfimages do even include additional info, such as width/height dimensions of the image, compression ratio, color space, bit depth, image encoding and resulting resolution as compared to the PDF page's own size:
kp@mbp:> pdfimages -list -f 3 -l 7 porsches-a4.pdf
page num type width height color comp bpc enc interp object ID x-ppi y-ppi size ratio
--------------------------------------------------------------------------------------------
3 0 image 1920 1440 rgb 3 8 jpeg no 20 0 175 175 182K 2.2%
4 1 image 1920 1440 rgb 3 8 jpeg no 26 0 175 175 130K 1.6%
5 2 image 1920 1440 rgb 3 8 jpeg no 32 0 175 175 92.1K 1.1%
6 3 image 1920 1440 rgb 3 8 jpeg no 38 0 175 175 233K 2.9%
7 4 image 1920 1440 rgb 3 8 jpeg no 44 0 175 175 238K 2.9%
To extract a specific page's images as JPEGs use the -j parameter:
kp@mbp:> pdfimages -j -f 11 -l 11 porsches-a4.pdf prefix
This would extract all images from page 11. Their names would be prefix-000.jpg, prefix-001.jpg, prefix-002.jpg etc.
NOTE: Sometimes direct extraction as JPEG isn't possible. pdfimages will still extract them, albeit in PNM or PPM format. You can easily convert these to PNG or JPEG by using ImageMagick's convert command:
convert some.ppm some.png
convert some.pnm some.jpg
If the images are still present, it is possible to extract them. You could use Photoshop or a tool like this: http://www.somepdf.com/some-pdf-image-extract.html
