Fast image processing with ExactImage

Searchable PDFs

The hocr2pdf program lets you create searchable PDFs from scanned originals. In principle, these PDFs consist of two levels: One is occupied by the scanned image, and a second  – often invisible underneath – contains the resulting text from OCR processing. Searching is done against this second level; it thus needs to match the graphical level as accurately as possible.

A number of programs use hocr2pdf to create PDF "sandwiches" of this kind. These include, for example, pdfocr, the largely automated pdfsandwich [4], xsane2sandwich  [5], and djvu2pdf [6], which extracts an hOCR file from DjVu data and creates a searchable PDF. How well this works is shown in Figure 3, in which the recognized words are highlighted by blue boxes.

Figure 3: A sandwich PDF generated by hocr2pdf.

Using hocr2pdf is relatively simple, although it does expect the hOCR data via the standard input channel (Listing 4, first line). You can generate the hOCR files with Cuneiform or Tesseract and edit them with the hOCR tools [7] if necessary. If the text alignment does not match the scanned image, this can be adjusted in hocr2pdf with the -s (--sloppy-text) option (Listing 4, line 2).

Listing 4

Using hocr2pdf


The hocr2pdf tool only supports a few options other than this; everything else is done automatically. The -i switch defines the graphical input file and -o the final output file. The -n option lets you prevent the integration of a graphical layer, and -r sets the resolution for it; hocr2pdf uses 300dpi by default.


For more complex transformations – for example, splitting a PDF file into individual images or comparing two images – you will still probably need to turn to ImageMagick or GraphicsMagick: ExactImage is not much help here. This also applies to identifying specific image formats; although edentify returns results, they do not stand comparison with the quality that identify will give you.

If you are more interested in the typical tasks of converting and editing popular formats, ExactImage offers a quick, easy-to-use alternative to the more complex tools in many cases. The hocr2pdf program proves to be a true gem that many other tools use under the hood.

Buy this article as PDF

Express-Checkout as PDF
Price $2.95
(incl. VAT)

Buy Linux Magazine

Get it on Google Play

US / Canada

Get it on Google Play

UK / Australia

Related content

comments powered by Disqus

Direct Download

Read full article as PDF:

Price $2.95