Published in Probe Volume 4(1-2): July 1993-July 1994
Bradley K. Sherman
Institute of Forest Genetics
USDA Forest Service
P.O. Box 245 Berkeley, CA 94701
bks@s27w007.pswfs.gov
`What is the use of a book', thought Alice, `without pictures or conversations?' --Lewis Carroll
High-speed digital networks and computers with graphical capabilities have made it possible to retrieve and view images almost as easily as one views purely textual material. This article is intended to be a quick introduction to this subject with some pointers to other resources.
What is a Digital Image?
Digitally stored images are analogous to a paint-by-number kit. The image is composed of rectangular regions, called pixels, and each pixel is assigned a number. Each number corresponds to a color or shade of grey. In order to reconstruct an image for viewing, one needs both the values for each pixel and a legend which maps the numerical values to a color. Image files on computers typically contain both the array of pixel values and the color map.
Digital images usually contain less information than a corresponding image captured with conventional photographic methods. Photographic film is an excellent repository for information. One 35mm slide can easily hold one hundred million bytes of data. This makes the digital storage of images somewhat problematic. Even if the cost of storage were not a factor, the time involved in retrieving and displaying very large stored images would be prohibitive. All common conversion of photographic data to digital data involves the loss of some information. Once the images are digitized, they are often compressed. Some compression methods result in further information loss. There is always a tradeoff between faithfulness of reproduction and amount of storage space required. Even if storage were free, it would be advantageous to keep the image size small for rapid transferal and viewing. This dynamic has led to many different formats for computer images, each with advantages and disadvantages. The decision about which format to use is highly dependent on the application, and on the hardware that will be used. On the Internet, exchanged images tend to follow two main formats: GIF and TIFF. These formats, particularly the latter, have many variations. JPEG images are a ubiquitous TIFF variant (JFIF).
In addition to these, different computer types can have their own internal formats, customized for their particular hardware. The PICT format on Macintosh is an example of this.
Software
There is software available for Unix, Macintosh and Intel-based platforms that will allow you to view downloaded images. You may retrieve them using anonymous ftp and experiment with them for free. Some of the software is shareware, and the author expects some small compensation if you like the software and continue to use it. (One author, for example, would like you to send a case of beer; another a postcard of your town.) Sources for image viewing software for major platforms will be found later in the article.
Hardware
Workstations such as those from Sun Microsystems or Silicon Graphics were developed with advanced graphical uses in mind. High-end Macintoshes and PC-clones are adequate, however. Monitor screens should be at least 15 inches, and graphics hardware may be helpful or necessary. Image analysis can be slow even on very fast computers.
Typical Image Sizes and Network Bandwidth
Images seen on the Internet can range in size from hundreds to millions of bytes. A four-megabyte file is not uncommon. Internet connections can be characterized by the bandwidth of the connection. Typical bandwidths are 9.6, 14.4, and 56 kilobaud. Major nodes may have one megabaud connections or better. Bytes per second can be approximated by dividing the baud rate by 10. A 56 kilobaud channel will transfer about 5600 bytes per second.
A typical 100 kilobyte image will take at least 1 second to move across a one megabaud channel and more than 17 seconds across one of 56 kilobaud. In addition, the presentation of the image on the computer screen may take seconds once the image has been transferred.
To the user, large images or slow network connections will be seen as delays. An occasional 10-second delay after pressing a key or mouse can be tedious in an interactive environment, arguing for very high speed network connections or using smaller image files. Compression techniques can be very useful, particularly when animated images are transferred.
Image Capture
There are at least two sorts of devices which one can use to digitize an image. Both make use of charge-coupled devices (CCD) which use quantum effects to transduce light to a pattern of electrical signals. Scanners have a linear array of CCDs which are mechanically moved across a flat image (in the manner of a xerographic machine). CCD cameras have a two-dimensional array of devices so that the entire image is captured at once. CCD cameras use conventional optics and hence can be used to take pictures of 3-D objects.
CCD devices are susceptible to thermal noise; they will produce small random signals even in absolute darkness. To increase signal-to-noise ratios, the devices can be cooled. Some CCD cameras come with refrigeration units for this purpose. CCD cameras are more expensive, but allow for a higher throughput in a production setting.
Commercial photographic processing labs commonly have equipment to directly digitize color slides. Copy stores will often have scanners available for rental. These work quite well with color or black-andwhite prints.
Where to get Image Viewing Software
All of this software may be retrieved using anonymous ftp.
Macintosh
JPEGView
ftp to sumex.aim.stanford.edu
directory: /info-mac/app/
NIH Image
ftp to zippy.nimh.nih.gov
directory: /pub/nih-image/
Unix/X11
xv
ftp to bongo.cc.utexas.edu
directory: /gifstuff/xwindows/
xloadimage
ftp to bongo.cc.utexas.edu
directory: /gifstuff/xwindows/
Intel/Windows3
Lview
ftp to oak.oakland.edu
directory: /pub/msdow/windows3/
Intel/MSDOS
cshow
ftp to bongo.cc.utexas.edu
directory: /gifstuff/ibmpc/
For More Information
These Usenet electronic conferences are sources of useful discussion:
alt.binaries.pictures.misc
comp.graphics
alt.graphics.pixutils
List of Frequently Asked Questions (FAQ), with answers, for these conferences are available by anonymous ftp from rtfm.mit.edu