IMAGE QUALITY
The digital capture process creates a digital object that should be considered a picture of the object. It exists as a whole image, that can be displayed by manipulating viewing software such at PhotoShop or Adobe Acrobat. Digital objects can be displayed in various measures of resolution such as pixels as in "pixels per inch" (ppi) or dots as in "dots per inch" (dpi) or lines, as in "lines per inch" (lpi) used for half-tones. Achieving a basic understanding of the technical issues relating to image quality is important. Several links to information that addresses image quality and technical standards are included in the resource section.
Decisions relating to image quality will depend upon the final outcome to be desired. Higher quality standards will be necessary if the outcome is to serve preservation purposes as well as access needs. Your choice of repository and the delivery method will affect your image quality choices. The condition of the original material, and choice of scanner affect image quality. The requirements for reproducing the digital object will affect the amount of image manipulation you choose to do. Scanning software usually comes with image manipulation software that allows the original photo to text page to be changed significantly. You will need to consider the purpose of your scanning project and balance the purpose with the need to reproduce the original as faithfully as possible. Certainly whenever you change or modify an image in any way you should note that change and report it in the documentation for the project.
Image enhancements
during processing:
Frequently source documents, photographs, stereographs, maps, and artwork
are in poor condition. Besides fragility, they may have lost contrast,
color definition, and clarity. In addition to physical remedies such as
conservator methods, electronic remedies are available. Tonal improvements
are possible after initial scanning into digital format. Contrast can
be increased to improve OCR accuracy. Software such as PhotoShop can be
used to enhance image clarity for the user. A decision will need to be
made regarding the amount of image enhancement that will be allowed. A
general statement clarifying how and why the images were manipulated should
be provided to the users. Examples of image enhancement - one,
two,
three
Image quality and
display:
Most personal computers display images in 100 dots per inch (dpi) or less.
A quality of display in excess of 200 dpi is often lost on present monitors.
If the object is scanned at higher resolution, it won't be displayed in
its entirety on a monitor screen. It must be either divided into sections
or 'panned' from top to bottom and side to side.
Multiple densities
(Thumbnail, medium, high resolutions)
An image can be captured at a high density and stored or displayed in
a variety of sizes and formats for different purposes. The "archival"
image can be scanned and stored off-line at 300 to 600 dpi Tagged Image
File Format (TIFF). From that other files can be derived and used. These
will be smaller images with lower densities that will load and display
much more quickly.
The smallest image, called a thumbnail, can be derived and used in the indexing display. When searchers obtain lists of objects matching their query, they must choose among items for further pursuit. This decision process is based on textual information, such as indexed metadata and visual cues from looking at the object. For images this presents a problem. The detailed images are large files. Download times over the Internet make this process frustrating for more than a few images at a time. Smaller images with lower file size and resolution are produced. These stamp-sized images, called thumbnails, are usually less than 10,000 bytes (10kb) and are quickly displayed. Often thumbnails are displayed at 72 dpi for quick loading.
For display of a digital collection on the Web, many sites offer a mid-range size and resolution option. These images will fill the viewer's monitor but are not more than 200-300 dpi in either a GIF or JPEG format. This is usually an acceptable level of legibility for general use. (Interactive page choose photos with problems, text with foxing, staining, uneven print, handwritten, outsized pages, three dimensional objects)
High-resolution images, above 500 dpi, can be produced for specific viewing. The difference in displayed resolution between a medium and high-resolution file is difficult for the general viewer to see. High-resolution image files are often available only upon specific request.
Differences between
archival and use files
Example Decisions on Digital Image Formats
http://sunsite.berkeley.edu/Imaging/formats.html
Capture Process
Processing the collection: The objects can be prepared for capture
by removing staples, unfolding materials, even enhancing faded originals
using conservatory techniques. Care must be taken to preserve the originals
such as removing bindings on books, inserting fragile papers in clear
retaining jackets, and unfolding pages. Once the material is ready for
capture, the methods of capture vary depending upon its form. Scanning
devices exist to capture slides, movies, maps, and documents. However,
the digitizing tools go further to include file conversion programs for
material already in digital form (e.g., word processing files, GILS data
layer files) but that require conversion to the target formats.
With the current generation of lower cost, desktop scanners, and the use of scanning and indexing software, and portable storage media such as small web servers, ZIP disks or recordable CDs, a local library or small organization can collect indexed images.
If an institution is searching for a permanent, archival copy of a collection, or microfilm, a combination of filming and scanning may be the best solution. If the objective of digitization includes archival preservation, then a durable medium such as microforms should be considered. High-resolution images of the objects can be rendered on the microform (usually microfilm). Since the patron will view the digitized object on a screen with limited resolution, capture of the image onto microforms creates the sharpest reproduction. Conversion to electronic image formats can be accomplished through scanning of the microform image. This process produces good results when trained microform camera operators and good equipment are employed. Future improvements to scanning quality can be utilized by rescanning from the microform. It is expected that ultimately electronic scanning will reach or exceed photographic quality. Perhaps the durability of electronic image files will also exceed that of today's microforms.
Other issues, beyond preservation, may justify the use of microforms as an intermediate process. Microform cameras are portable, require minimal calibration, and can be operated by local, trained operators. This allows volunteers, willing to receive camera operation training, the opportunity to process objects on site. The film is then sent to a central service for processing and scanning. The electronic files are often loaded onto CDs or high capacity disks (e.g., Iomega's ZIP disks) and returned to the collection site for indexing.
If microforms are
not desired, then scanning the objects directly into electronic files
is possible. Typically, the objects are scanned at the highest resolution
to create the master image. Lower resolution image files are then made
from the master for medium quality and thumbnail-size images.
OCR versus text images
Optical Character Recognition processing can be applied to textual materials scanned into digital image files. The software will convert patterns of light and dark on the image to characters suitable for word processing viewing. However, even with excellent quality textual documents, some errors in interpretation occur. With decreased contrast, clarity, and typeface size of the source document, read errors increase. Additional human interpretation is required. Typically, this is done on the computer screen, displaying the recorded image and asking the operator to assign the proper character. Even with a spell check, errors in interpretation will get through validation steps. If the purpose of the OCR is to create a full-text reproduction of the words on the source document, then substantial human assistance is required. However, if the objective is only to create a concordance file of words from the document for search and retrieval, then OCR may be cost-effective.
- Save digital masters in TIFF, display digital objects in JPEG.
- Establish measures for image quality standards that you adhere to consistently.
- Establish routines for reviewing images for quality and legibility.
- Hire staff whenever possible who have experience with photographic images, or photographic techniques.
- Decide how you will display captured digital objects before you begin scanning. Most scanning decisions relating to quality, file size or resolution will directly relate to the display image you want to create.
- Study the resource
documents before you begin your project.
