Keynote Papers
Trends in Internet Imaging and Implications for Spatial Applications
Carey Mann
Chief Operating Officer HMR Inc., 10420 Little Patuxent Parkway, Suite 470, Columbia, Maryland, USA 21044
Tel: 410-730-4330 Fax: 410-730-4359
Email: carey.mann@hmrinc.com
Key Words Imaging; Internet; Image Server
Abstract The Internet is changing everything and imaging is no exception. While the last five years have witnessed an impressive growth in the use of aerial and satellite imagery, the real explosive growth that the Internet will facilitate is only just beginning. While the inherent value of imagery has not changed substantively in this period, the audience demanding imagery has broadened, the nature of supply has shifted from contract photography to data merchandising, and the enabling technologies in digital and Internet imaging have removed many practical impediments to access and utilization. This paper reviews the state of digital imaging, emerging technologies for Internet imaging, trends in imagery data acquisition and the implications for GIS and Mapping professionals.
On the Cusp of an Imaging Revolution:
The disruptive technologies of the past two decades testify to the rapid pace of change as we begin the new millenium. The PC has changed the way we live and do business, and has created many of the world's leading companies. The Internet revolution is only just beginning, yet in its five-year history the Internet already has a similar story. It is numbing to think that Microsoft, Intel, Dell, and Oracle did not exist twenty years ago. AOL, Amazon.com, and Yahoo have yet to turn ten.
Many believe that the digital camera, and all that it portends, represents the same type of disruptive advance. Shipments of digital cameras are expected to reach 3.6 million units in 1999 and grow at an annual rate of at least 36% over the next three years. This estimate may be conservative; growth will likely be exponential once quality is improved, cost reduced, and the user experience simplified.
The digital camera, digital image, and Internet are entirely complimentary and intertwined technologies. The Internet is highly visual and only bandwidth restricts more extensive and creative use of images by businesses than is already seen today.
Considering consumers, InfoTrends notes that in 1999 approximately 25% of U.S. households owned a scanner. The majority of these users utilize community, film processor, photo sharing, photo portals, and vendor Internet sites to share photos. In 1999, they estimate users of online Web album sites shared photos with nearly 100 million viewers.
Dataquest predicts that the market for consumer photo editing software will grow from a mere $300,000 in 1995 to well over $400 million by 2002. Even for a market in its infancy, these numbers are staggering.
But, even so, this activity only scratches the surface. Kodak estimates that 82 billion pictures are taken each year adding to an already immense personal and professional archive. Clearly, to realize the full potential of products and services related to the digital image requires an imaging infrastructure integrated into, and architected for, the Internet.
The mass adoption of the digital camera will have a profound effect on the way images are captured, shared and used. Many believe that within five years film will be all but obsolete, and that the digital camera will become ubiquitous. It will also take many forms beyond the traditional hand-held camera, such as hybrid video/still cameras, Internet cameras and embedded devices. The ease of image capture will fuel intense demand from consumers and businesses for Internet enabled imaging technologies, products and services.
Taking Stock of the State of the Art
The Internet has changed everything. Yet, the medium is crude by any measure. As HTML gives way to XML, as base bandwidth expands and as the infrastructure is improved, the focus will turn away from crude mimicking of existing mediums to the richness of the experience and the integrated and dynamic use of imagery, voice, video, and other data.
The digital camera will change everything about how images are captured and used, and the Internet will be an integral factor to its evolution. Like any disruptive technology, the initial enthusiasm gives way to diminished expectations as the reality of removing technical impediments sets in. This momentary pause obscures the market dynamics and results in a gross underestimation of the impact of the technology. We are at about this point with the digital camera.
The digital camera will make film based cameras obsolete. Every family will own one, and businesses will use them with greater frequency than was ever the case with analog cameras. To gain significant momentum, two things must occur. First, digital cameras superior and easier to use than 35mm cameras must sell at similar or lower price points; this inflexion point could be reached as early as 2001. Second, the consumer experience for digital images needs to improve dramatically, including home printing and retail and Internet photo services; like Internet bandwidth, the incentive and inertia to solve these issues will be driven by the demand created by volume adoption of the cameras.
This emerging market will be huge, and it will create demand for imaging software of all types. It will be difficult to distinguish the infrastructure from the client solutions; they must work together to realize the full potential of digital imaging over the Internet.
Imaging Market Trends:
Any current quantification of the market for digital imaging and Internet imaging software will obscure the dynamic and explosive growth that can be anticipated as the Internet matures and the digital imaging revolution unfolds.
According to Michael Foss, General Manager of Consumer Digitization and Vice President of consumer Imaging at Kodak, of the 82 billion pictures taken every year, only 2% of these pictures are ever reprinted or reused. Foss believes that the opportunity with digital images is not only to convert some of this massive volume and historical inventory of analog prints to digital form, but also to increase the usage rate at which images are reused or repurposed.
This transition from film-based photography to digital photography is already underway worldwide. According to InfoTrends Research Group, Inc.'s 1999 Low End Digital Camera Forecast Worldwide Summary, "Though the complete transition is still many years away, it is happening in all market segments, including youth, consumer, and business.
InfoTrends believes that "online photo sharing is the next hot Internet application" and expects the number of sites and volume of photos to grow exponentially. They state that the essential factor to the success of each of these sites is "an adequate technical infrastructure not only at the time of launch, but as the number of viewers and stored photos grows".
Internet Imaging Technology
Three important factors are driving Internet Imaging technology: 1) Bandwidth; 2) Server side processing; and 3) Image compression. No factor is independent of the other.
The bandwidth problem is being solved in two ways. The first solution is obvious - improvements in bandwidth capacity itself. Digital Subscriber Lines, cable, and satellite services are lowering the cost and improving baseline bandwidths for consumers and businesses alike. The second solution is also obvious - use less bandwidth. The most promising method to reduce image bandwidth consumption is the Internet Imaging Protocol (IIP). IIP, a data transport standard supported by computing leaders such as Microsoft, Kodak and HP, dramatically improves the Internet experience for users of large or high-resolution images. IIP provides an infrastructure that efficiently manages what image data is sent to a browser or remote application and improves performance through image compression, caching and other methods.
Image compression, although a subset of an Internet imaging infrastructure, is critical as it dramatically reduces bandwidth requirements for a given image resolution. Typically, JPEG, because of its ubiquitous presence on the Web, has been the compression format of choice.
Nonetheless, Wavelet, which offers superior compression at a given quality of image than JPEG and is decoded in an Internet friendly manner, has been gaining favor. Wavelet can achieve lossless or near lossless compression ratios of 20:1 or better. When compressed wavelets are decoded, the data is retrieved from the file in a manner that is similar to streaming. By rendering a Wavelet file over the Internet, Wavelet vendors are able to deliver the data needed to view a given area at a specified resolution without requiring the user to download the entire file. Wavelet, however, requires proprietary plug-ins and file conversion, and these factors have focused its growth on certain vertical markets.
JPEG 2000, which incorporates Wavelet compression into the standard JPEG format and is an industry wide standards effort, will likely be the dominant format and the best vehicle for widespread acceptance of Wavelet technology. It will build upon the JPEG legacy and allow Wavelet to be viewed royalty free in any off-the-shelf browser, the critical factor that will ensure its broadest adoption. We should see it emerge and become popular as early as 2001.
Server side processing adds scalability and robustness and automates server based image display, processing, printing, and other imaging functions. An Internet server implies a distributed multi-tier architecture, where tasks are split more efficiently between server and client and where data resides where it can be most efficiently utilized.
Implied by an IIP server architecture is a non-destructive editing model, where the client instructs the server how it wishes to manipulate a portion of the image and the server returns the pixels for that region appropriately manipulated. This editing model not only differs from traditional implementation that require the entire file to be downloaded to be edited, but also lends itself to reduced data redundancy, better data management, and more user flexibility. The source file need not be replicated for each user edit, as the edits can be replicated on the fly by the server from commands stored at and sent by client applications.
While all of this efficiency is all very nice, why is it important? It is important because offloading the processing to a server enables users to have rich and simple to use client applications using standard browsers or easily distributable plug-ins. It is important because the distribution of the application to users is automated. It is important because it facilitates widespread access to vast of amounts of data.
The implication, however, of server side image processing is much more significant than simple computing efficiency. The potential to reach mass audiences with easy to use applications is what is most compelling.
The Democratization of Imagery Acquisition
Internet based imaging solutions will be increasingly popular for GIS and engineering applications. Transportation authorities, utilities, resource companies and local, regional and national government agencies are heavy users of aerial photography. Moreover, there are large archives of scanned large format documents such as maps and engineering drawings at most such organizations. Finally, the digital camera is fast becoming the method of choice for documenting the existence and condition of physical assets.
The growing demand for better methods to disseminate such data within these organizations and legislation in many countries that mandates government agencies to make data more widely available for public use will drive this demand.
So too will the new breed of satellite imaging companies like Space Imaging, Orbimage and EarthWatch. These companies have recently launched satellites that will enable them to sell 1-meter high-resolution on-demand imagery at very cost effective prices.
Aerial photography companies, such as the Canadian firm Hauts-Monts in their CitiPix joint venture with Kodak, are also now providing imagery on spec. CitiPix imagery will cover an extended radius of 95 North American cities with ground resolution of six-inch pixels or better, and will be available to purchasers this year.
Obtaining such imagery had previously required expensive contract photography, long lead times and intricate planning. Now, users now will be able to obtain their high-resolution imagery from the Internet, as they need it, and relatively inexpensively.
While on the surface, this would seem to be a supply driven market, the opposite is in fact true. The demand has always been there, but the cost, timeliness, and availability of the imagery had precluded most users from pursuing it. The Internet, together with satellite and aerial acquisition capabilities, will untether this pent-up demand.
The imagery will find its way - and quickly - into every type of analysis and decision support software; the applications are only limited by the imagination.
Internet Imaging for Engineering and GIS:
Most things GIS and engineering are heading to the Web. Major software vendors in this area offer Web publishing solutions that permit both vector and image data to be published over the Internet.
But, these are only crude indications of where we are heading. Increasingly we will move away from a model of localized workstation solutions run by "analysts". Distributed computing, readily available off-the-shelf data, and integration with other enterprise business systems will characterize the new order.
If GIS as an island is dying, then so too are technological solutions that while pragmatic are proprietary and inward looking.
To take advantage of this new wealth of imagery will require a technological base built for the extended enterprise. The market dynamics that will be unleashed with digital photography and the Internet will further accentuate this requirement, and with every consumer and business likely to be immersed in such technology it is highly unlikely that less than state of the art will be accepted from GIS and engineering vendors.
Today, most GIS and engineering software vendors are only beginning to embrace enterprise computing, and especially so with imaging technology. In imagery, typical strategies are confined to reducing the amount of image data, either through image compression or by subsampling the image at a lower resolution. Neither solution is satisfactory as compression alone is a subset of what is required in an efficient Internet imaging infrastructure and images of degraded quality do not meet the need of most applications.
The most widely accepted "data reduction" method is Wavelet compression, such as the solution offered by LizardTech. The compression results can be dramatic, but the user must adopt a proprietary format and accept the compatibility and interoperability problems that invariably follow from doing so.
And, compression, even when delivered over the Internet from a "server" is only one element in an Internet Imaging Infrastructure. Nonetheless, it is illustrative of what is possible and what is coming.
Bentley System's affiliate HMR has developed a comprehensive IIP image server that is designed for enterprise implementations, and it may perhaps be the most ambitious Internet imaging technology yet developed for any market.
This technology is incorporated into Bentley's MicroStation and ModelServer products, and offered by itself under the brand name ModelServer Imager. ModelServer Imager is a high performance IIP image server that supports a variety of image formats and compression codecs natively. Wavelet, for example, is but one codec supported, and JPEG 2000 will be supported, as it becomes available.
Client side development options include HTML rendering, auto-downloadable Java applets, and ActiveX controls. The server features a plug-in architecture for server side processing and for image format support.
The architecture features a non-destructive editing model, which executes edits at display dynamically through commands, rather than by permanently altering the pixels. Since the server can send commands rather than pixels, extensive editing of very large files is possible even over the Internet. Such server side processing can be automated from a browser event using a scripting language called PictureScript.
On line demos of ModelServer Imager can be seen at
www.hmrinc.com.
In Conclusion:
The use of imagery in GIS and Engineering will change dramatically as the Internet's potential is embraced, and as the digital imaging revolution unfolds. Widely available high-resolution imagery at affordable prices will extend access well beyond GIS and engineering specialists to a wide array of planners, analysts, and managers in government and private enterprise. And it will be mainstream imaging technologies reapplied for spatial applications, which will make all of this possible.
References
Contained herein.
