Scanning Artifacts: Using a Flatbed Scanner to Image Three-Dimensional Objects
Brett A. Houk and Bruce K. Moses
The articles published in the technology column of the SAA Bulletin often present archaeological applications of new technologies that require extensive training or the purchase of new equipment or software. In this article, we describe a new application for an existing technology that is relatively inexpensive and already in widespread use. Many archaeologists and technical illustrators use flatbed scanners to import profile drawings or plan maps, and then redraft the maps using a computer drawing program. Over the past year, the staff of the Center for Archaeological Research (CAR) at the University of Texas at San Antonio (UTSA) has been experimenting with using the same flatbed scanner technology to scan artifacts, rather than photographing or illustrating them. While we are aware that others are doing this as well, we have not yet seen any discussion of the method in the archaeological literature. Here we discuss the scanning methods we have developed, as well as the advantages and the limitations of using this technology.
To demonstrate the versatility of this technique, we present several examples. The first is a chert dart point. The second is a cow rib with butchery marks made with a metal tool. The latter demonstrates the ability of the scanner to accurately image larger three-dimensional objects. We also present a series of scanned images of different artifact types as examples of the versatility of the technology. In most cases, the process involves several steps: scanning the artifact to the appropriate scale, adjusting the image using DeskScan II, saving the image as a computer file, and manipulating the image using Photoshop LE.
Before scanning an artifact, several important decisions must be made. The resolution at which the image will be scanned and the type of image to be produced (i.e., color or black-and-white) must be selected. Generally, the end result of the process guides these two decisions. As discussed below, file size increases with resolution. Therefore, computer memory limitations and the final output resolution (i.e., for Internet publishing at low resolution or printing in a technical report at high resolution) generally dictate the scanner settings.
The HP DeskScan II software allows the user to set the image type, print path, and scaling before scanning. The image type menu includes items such as black and white photo, sharp black and white photo, color photo, color photo, sharp color photo, millions of colors, and sharp millions of colors, along with a variety of halftone and line art options. The print path menu allows you to select the dpi above which the image will be scanned (Hewlett-Packard 1995b, HP DeskScan II User's Guide. Publication No. C2522-90002. Hewlett-Packard, Palo Alto, Calif.). The image may be scaled before scanning as discussed.
Our first example is a Pedernales point from Central Texas. This point is made of a brown, fine-grained chert. We have included a drawing of the artifact for comparative purposes (Figure 1a). The first step is to place a clear piece of plastic on the surface of the scanner bed. This prevents sharp objects from scratching the scanner's glass plate. We use a sheet of blank overhead transparency film. The artifact is then placed on the plastic and the lid to the scanner gently closed to hold the artifact still. Depending on the thickness of the artifact, however, it may be impossible to close the lid. Leaving the lid open results in a black background around the object.
Our example was scanned at 300 dpi, as a sharp black and white photograph, at 100 percent. To scan the image, we first used the preview option on the DeskScan II software and then selected the final area we wanted to scan. We also used the automatic exposure option to adjust the brightness and contrast of the image. This was done by first selecting a section of the preview image of the projectile point and a small section of the gray background and then clicking the automatic exposure button. The contrast and brightness also may be manipulated individually, but we have found that the automatic exposure is fairly consistent and produces a good image. For final scanning, we selected an area just slightly larger than the projectile point to minimize file size. The image was then saved as a Tag Image File Format (TIFF) file directly from the DeskScan II program (Figure 1b). The area around the artifact appears as a light gray background because the artifact prevents the scanner's white lid from contacting the flatbed surface. Had the artifact been thicker, the background would have been darker.
We then used Adobe Photoshop LE to open the TIFF. In Photoshop, we used the eraser tool to remove the gray background around the point (Figure 1c). The file was then saved, again as a TIFF in this case. Other file formats should be considered depending on the intended use of the image. For example, both Joint Photographic Experts Group (JPEG) files and Graphics Interchange Format (GIF) files are suitable for publishing on the Internet.
Our second example is a fragment of a cow rib from 41BX437, a Spanish colonial site associated with the Alamo in downtown San Antonio. This piece demonstrates the depth to which the flatbed scanner can "see" a three-dimensional artifact, the utility of zooming in on a section of an artifact using the scaling feature, and the ability to annotate the image using Photoshop. The particular piece of bone we selected measures 265 mm long by 55 wide by 14 mm thick (Figure 2). This object was scanned at 75 percent actual size as a sharp black and white photograph at 300 dpi, with the automatic exposure option.
As is shown in Figure 2, the scanner does an excellent job of imaging an artifact of this size. The sharpness of the image decreases with distance, but detail is still discernible as far as 18 mm. This file was also saved as a TIFF.
An extremely useful feature of the scanner is its ability to scale an image prior to scanning it, thereby maintaining the desired resolution. As an example, we zoomed in on the area of the bone marked by the dotted line in Figure 2 to produce Figure 3. Figure 3 was scanned with a path of 300 dpi at 300 percent. The contrast and brightness were then adjusted manually to highlight the indicated butcher marks.
This test of the scanner demonstrates clearly the decrease in sharpness and the increase in darkness that occur with distance. The scanner does a good job of recording the ruler to a distance of approximately 12 mm. Between approximately 12 and 20 mm, the text is still easily read, but the image is darker and the lines begin to lose their sharpness. After about 27 mm, the image becomes very dark, the text illegible, and the lines fuzzy.
To test the effects of increasing an image's scanned resolution, we used the projectile point from our first example and scanned it at 50 percent normal size at increments of 100 dpi (Figure 5). In all cases, the artifact was scanned as a sharp black and white photograph with constant brightness and contrast levels and saved as a TIFF. At 100 dpi, the image required only 18 kilobytes (k) of disk space to save. At 600 dpi, it required 660k. The original full-size image required 660k at 300 dpi. Figure 5 demonstrates the differences in image quality and file size at different resolutions.
One distinct advantage that high dpi images have, however, is that it is possible to resample the image's resolution downward without decreasing its length or width. For example, a 4-x-6-inch, 600 dpi image can be converted to a 300 dpi image of the same dimensions using Photoshop or a similar software package, but a 300 dpi image cannot be resampled to a 600 dpi image without a proportionate reduction in its dimensions. Similarly, a 600 dpi image can be enlarged to show detail and still retain a high resolution, as is illustrated in Figure 5.
The scanner cannot be used to produce images of very large or heavy artifacts. The maximum scanning area is 216 x 356 mm (HP 1995a). We could not find a maximum weight that the scanner bed can support in the HP documentation, and we decided not to test it ourselves. We do not recommend trying to scan heavy artifacts such as metates, large celts, or carved stone monuments!
The process has important advantages over photographing or illustrating artifacts. It is relatively quick and can be completed entirely in house. More importantly, the quality of the final image is immediately known, unlike artifacts shot on film that must be processed. The quality of the images, while dependent on the various factors discussed above, is generally very high. Because the images are already in digital form, they can be manipulated and placed into manuscripts easily. Perhaps the most important advantage is cost. A scanned image costs very little to produce; essentially, once the scanner and necessary software have been purchased, the only cost is the time of the person doing the scanning. "Reprints" of the image as computer files or printed copies are also virtually free and easily made.
We have included several examples of the range of artifacts that we have scanned in Figure 6. Some of these images have had the background erased, while others have not. Each of these examples was scanned at 300 dpi as a sharp black and white photograph. It is possible to add a different background to an image using Photoshop or a similar software package. This is particularly useful for color images published online.
Students faced with trying to produce a high-quality dissertation or thesis at a low cost should find this technology extremely helpful. Professionals should benefit from its versatility and cost effectiveness as a tool for archiving images, sharing data over the Internet, and producing technical reports. While we have only been experimenting with this technology for about a year and our techniques are not completely refined, we are extremely pleased with the results thus far.
Brett A. Houk and Bruce K. Moses are both at the Center for Archaeological Research at the University of Texas, San Antonio.