New Deal Stage | Back to the Overview

Summary: High Quality Imaging for Pictorial Items and Documents with High Artifactual Value: The IBM Digital Library Demonstration

Introduction. The IBM Digital Library Demonstration included two interrelated features: (1) a very high resolution digital camera and (2) an associated database for managing the stored files. The project is described in more detail in a final report.

Initial Capture Images. The IBM activity created very high resolution images of selected pictorial items from the collection, as well as documents of special significance or those presenting special challenges in reproduction. The first step in the production process was the IBM-Research-developed Pro/3000 Scanner and an IBM PS/2 workstation, equipped with the IBM-Research-developed PISA95 scanning application. The Pro/3000 is an overhead device, occupying the position that a film camera might occupy on a copy stand. The PISA95 application supports a two-display configuration that allows the user to control the program on one display, while presenting high-quality images on a second monitor. The Pro/3000 had been previously used to digitize materials in the Andrew Wyeth Collection, and in the collections of the Vatican Library and the National Gallery of Art.

The Pro/3000 scanner produced archival quality, high resolution color images from negatives and from reflective media in the collection. It is based on an IBM-proprietary charge-coupled device (CCD) imaging sensor chip that provides exceptional 12-bit dynamic range and superior noise performance. The system employs special colormetric filters that enable the camera to capture a broader spectrum of colors than most digital scanners. Typical scanners have color sensitivities that cater to known color emphases of photographic emulsions but may not be equally sensitive to the full range of colors found in nature or works of art. In order to prevent color contamination, the Pro/3000 must be used in a room with black walls and ceiling.

The CCD sensor moves across the image plane internally. This eliminates the need to move the camera or the media being scanned. The Pro/3000's digital camera is supported by a motorized column. A bellows is used to focus the digital camera. Because of the limited height of the column, the largest original that can be scanned is 45 by 60 cm. The system lighting employs dichroic filters that reflect heat away from the objects being scanned. A special book holder has several protective features, including hand control over pressure applied to pages.

The capture images have spatial resolutions that range as high as 3072x4000 pixels. As the images are captured, they have a tonal resolution of 12 bits per pixel in each color channel (red, green, and blue), or 36 bits per pixel overall. After review, the images are saved as uncompressed 8-bit-per-channel (24 bit) files. The image format includes certain elements unique to this IBM system; the file contains a TIFF header. The images range in size from 5 to 21 megabytes.

The PISA application works with the Pro/3000 to capture monochrome or color images. Its user interface displays both the proper digital camera height and the bellows position for proper focus for a number of sizes of originals, helping the scanner operator to more quickly position and focus the digital camera when the size of the original is changes. PISA also includes a utility that records the lighting pattern produced by the illumination. In the process of creating a scanned image PISA uses this information to correct the image, so that it appears as it would if the illumination were spatially uniform. PISA also includes a utility to analyze the scan of a color-calibrated test chart and determine the color characteristics of the scanner on the basis of that analysis. In the process of producing a scanned image, PISA uses this information to correct the colors of the scanned image so that they will appear correct.

Very large items, e.g. billboard sheets, were captured in segments. A typical sheet, for example, might be captured in a set of four to nine images, to be concatenated in later stage of the process.

Image capture was carried out by a pair of Library of Congress staff members. The items to be scanned were brought to the room, removed from their housings (an acid-free storage folder and, in some cases, a mylar sleeve) and placed on the camera stand. If the new item to be scanned differed in size or reflectivity from its predecessor, one of the operators would adjust the camera's height, framing, and focus. Next the 12-bit-per-channel exposure was made and reviewed, and data was entered in the database. Then the system automatically reduced the tonal depth to 8-bits-per-channel and the file was saved on the capture workstation's hard disk and the original item was returned to its housing. In a separate process, batches of saved images were transmitted to a server via a local area network.

Including setup, review, and recording to media, each grayscale exposure took about 2 minutes while color exposures took about 8 minutes. A typical work session lasted for about 6 hours and saw the capture of 80-90 images. Scanning was carried out on a part-time basis from December 1995 to January 1997. During that period, 3,000 images were captured.

Derivative images. After being saved to a server, the initial-capture images were processed to produce an archival image and a set of additional images to serve access needs. The derivative images were produced in a batch mode by Library of Congress staff members using IBM's Comit95 software. Approximately 300 archival images constituted a batch; the process ran unattended. Each batch required about 8 hours to process, including the time-consuming movement of the source images over the local area network and a visual check of all thumbnails.

The archival image is more or less of a copy of the initial capture, regularized to eliminate the special IBM features.

The three service images included in this online presentation were also derived from the initial capture images. In order to provide more efficient service via the World Wide Web, the files have been compressed and two of the three have been reduced in scale.

Image-management database. As image capture proceeded, the Library team entered data about each item into a relational database. This database was associated with the IBM Image Database Subsystem (IDBS), implemented on two IBM RISC System/6000 workstations running the AIX operating system. This subsystem includes two of IBM Digital Library solution components: the VisualInfo Library Server, and the VisualInfo Object Server.

Some of the database fields contain data needed to manage the storage of the digital files, the type of information sometimes called "administrative and structural metadata." Other fields contain descriptive information about the particular image begin captured. Multi-page documents were captured one page at a time; the descriptive information, therefore, was recorded at the "page" level.

Click here for a list of the database fields.

New Deal Stage | Back to the Overview