Previous | Table of Contents | Next |
The computation of data storage can be illustrated through the example of using 3.5 × 5 inch photographs in a visual database. A scanner is used to digitize each photograph using a resolution of 300 dots per inch. Then, without considering the effect of the color depth of the scanned images, each photograph would require 3.5 in. × 5.0 in. × 300 × 300/(8 bits/byte) or 196,875 bytes of storage.
Exhibit 2 compares the data storage requirements for a 3.5 × 5 inch photograph scanned at 300 dpi using different color depths.
In examining the data storage required for the 3.5 × 5 inch photograph, note that the maximum number of colors supported based on the indicated color depth is indicated in parentheses to the right of the color depth. As indicated by the entries in the table, the use of color significantly affects the data storage required for an image. This effect governs not only the number of images that can be stored on a visual database, but also the ability of users to work with stored images. Concerning the latter, the memory requirement of a workstation to display an image is proportional to the amount of storage the image requires. Thus, physically large images scanned at a high resolution with a high color depth may not be viewable, or only partially viewable at a time, on many workstations.
Careful selection of a color depth appropriate to the particular image-based application supported can result in significant savings in data storage and the time required to transmit images. For example, in a personnel database, a color depth of one or a few bits would probably be sufficient for pictures of employees. Note from the preceding table that the use of black and white images requires 196,875 bytes to store a 3.5 × 5 image, whereas the use of a 24-bit color depth results in a data storage requirement approximately 24 times greater. This means that the use of black and white pictures of employees could reduce the data storage requirements of a personnel database containing employee images by a factor of 24. For an organization with hundreds or thousands of employees, these savings could translate into a significant amount of disk storage becoming available for other applications, or the reduction in equipment required to support an image-based application.
Although few image applications use black and white, significant savings in data storage and transmission time can be achieved by selecting an appropriate color and color depth. In the real estate field, for example, the use of digital cameras is expanding. Many real estate professionals now take pictures of their listings and enter them into a central database for viewing by clients or other members of the organization. Although most digital cameras can support a 24-bit color depth, use of that color depth does not provide any appreciable viewing difference of a home, room, or swimming pool over the use of an 8-bit color depth. Thus, selecting an 8-bit color depth can reduce data storage of color images by a factor of three from the default 24-bit color depth used by many digital cameras.
Images also affect client/server operations because of transmission time. To illustrate the effect of image storage on transmission time, the example of a 3.5 × 5 inch photograph stored on a server connected to a 10Base-T Ethernet local area network (LAN) is used. There are 39 workstations and one server connected to the LAN for a total of 40 network devices.
Ethernet is a shared access network, meaning that at any one time only one user can transmit or receive information. Thus, although each device can transmit or receive data at 10M bps, on the average, the devices obtain the use of 10M bps/40 or a 250K-bps data transmission capability. Assuming the photographs were stored using a 24-bit color depth, then each image would require 4.75M bytes of data storage. Storage would actually be slightly more than that amount because images are stored using special file formats that add between a few hundred to approximately a thousand bytes of overhead to each file. Using a storage of 4.75M bytes and an average transmission capability of 250K bps, the time required to download the image from a server to a client workstation would be 4.75M bytes × 8 bits per byte/250K bps or 152 seconds. Thus, on the average it would take almost 2.5 minutes to download each photograph.
Displaying the image on a monitor would result in a slight increase in time because the server would have to access and retrieve the file containing the image from its disk storage system, and an image display program operating on the workstation would require some time to display the image. In any event, 2.5 minutes is not an appropriate waiting period for a guard attempting to verify personnel entering a building, a real estate broker attempting to show a client a series of pictures of homes, or a doctor attempting to view a previously performed MRI scan of a patient.
Several techniques have been developed to manage the storage and transmission challenges associated with the use of images in a client/server environment. They are presented in the sections that follow.
One of the most effective methods for reducing image storage requirements and transmission times is to use an appropriate file format when storing the images. Today, most imaging programs support a wide range of raster file formats, such as the CompuServe GIF, Joint Photographic Experts Groups JPG, Truevisions TGA, Aldus Corp.s TIF, and Microsoft Windowss BMP. Some of those file formats store scanned images on a pixel by pixel basis, using one to three bytes per pixel for storing the color depth. Other file formats include a built-in data compression feature that compresses the scanned image before storing it.
Although it is tempting to select a file format with a built-in compression capability to reduce data storage and the transmission time required to move images from a server to client workstations, the various compression methods have important differences. For example, CompuServes GIF file format uses the Lempel Ziv Welch (LZW) lossless compression method, whereas Aldus Corp.s TIF file format supports six different types of compression, including the Joint Photographic Experts Groups JPG lossy compression method.
Previous | Table of Contents | Next |