Project Aon

This How-to covers two OCR methods using different image capture hardware:

OCR with a flatbed scanner

The simplest OCR method makes use of a flatbed scanner. Most scanners now come with OCR software, though the quality of this software varies.

  1. Make sure that your scanner and OCR software is working properly. OCR software is used to scan in printed text into a text file.
  2. Place the book on the scanner. Scanning is usually performed on a double page spread; open the book to the spread you wish to scan and place it face down on the scanning plate.
  3. Scan in all of the text. You may need to refer to the instructions which came with your scanner and OCR software at this point.
  4. Repeat the process until all the the pages in the book have been scanned and recognised.

OCR with a digital camera

The disadvantage of using flatbed scanners is the need to flatten the book against the scanning surface for best results. There is an alternative OCR process which involves using a digital camera to photograph the text prior to the text recognition process. OCR software is still required and must allow recognition of text in image files.

  1. Prepare your camera. Most modern autofocus compact digital cameras should be sufficient for text recognition. Anything above 1.3 megapixels should be sufficient if a low compression "fine" mode is used, though Macro mode is essential for reliable results and should be turned on (this is usually indicated by a "flower" icon).
  2. Prepare your work area. Unless you are able to operate your camera one-handed, you will need something to hold the book open while you photograph the pages. For best results pictures should be taken in strong natural light with the flash turned off (the ink used in printing tends to reflect much of the flash, reducing the image quality). If this is not possible, try angling the book slightly while taking the picture to reduce the reflection.
  3. Photograph the pages. The amount of text you can photograph in one shot depends entirely upon the camera resolution and resulting image quality. In high quality conditions you may be able to shoot an entire page at once. However, in low quality conditions you may have to photograph each paragraph separately. To improve the OCR results, try to keep the camera straight and level while shootings. If possible, check each image on the LCD prior to moving onto the next. That way blurred images can be deleted and retaken while keeping the order of the images intact.
  4. Prepare the images. If required, rotate the pages so the text is in the correct orientation. At this point it is also useful to crop any unnecessary space around the text, this will reduce the workload of the OCR software.
  5. Perform OCR. Depending upon the OCR software being used, you will either have to process each image file separately or will be able to create a batch job to process all the images at once. If a batch process is used, ensure that the pages are in the correct order prior to starting the OCR process.