Digital Imaging Tutorial - Image Creation

6B. Technical Infrastructure: IMAGE CREATION

Key Concepts

introduction
how scanners work
scanner types
image processing

FILE/IMAGE PROCESSING
A variety of processing steps follow scanning. Such procedures may occur at any point in the digitization chain, from immediately after scanning to just prior to delivery to end-users. These may be customized modifications that affect only certain files, or mass, automated processing of all files (batch processing). They may be one-time operations or done repeatedly on an as-needed basis.

Examples of file/image processing operations:

Editing, touch-up, enhancement—this includes steps such as descreening, despeckling, deskewing, sharpening, use of custom filters, and bit-depth adjustment. In some cases, the scanning software performs these steps. In others, separate image-editing tools (e.g., Adobe Photoshop, Corel Photo-Paint, ImageMagick) are utilized.

Compression—sometimes carried out by dedicated scanner firmware or dedicated hardware in the computer. Compression can also be a software-only operation though dedicated hardware is faster and should be considered when creating very large files or very large numbers of files.

File format conversion—the original scan may not be in a format suitable for all intended uses, thus requiring conversion. See Presentation.

Scaling—it's likely that scans captured at high resolution will not be suitable for on-screen display. Scaling (that is, resolution reduction through bit disposal) is often necessary in order to create images for Web delivery. See Presentation.

OCR (optical character recognition)—conversion of scanned text to machine-readable text that can be searched or indexed.

Metadata creation—addition of text that helps describe, track, organize, or maintain an image.

Computer Considerations
In some cases, image processing can be accommodated in the scanning workstation, especially if each image is checked as it's created. In the case of "on-the-fly" operations such as image scaling done just prior to delivery, image processing usually takes place on the image server.

Other operations may call for a separate computer. Image editing, especially for uncompressed 24-bit color images, requires large amounts of RAM and video memory. To work most efficiently, image editors require RAM several times the uncompressed size of the file being edited. A large, high-resolution monitor is also needed.

Image processing steps that may be carried out on every file (e.g. OCR, format conversion, deskewing) can be extremely CPU intensive. Batch processing requires a fast processor, lots of RAM, fast storage subsystems, and rapid and efficient routing of data within the system. These characteristics are more often found on multi-user systems. In particular, Unix systems, with their inherent batch processing capabilities, are well-suited for these kinds of tasks, though computers running Linux or Windows 2000 Professional or XP Professional may also be suitable.