Revisiting OCR As An Auto ID Technology
For many of Dade Behring's customers, next-day delivery is not good enough. The $1.3 billion laboratory instrument manufacturer often has to ensure same-day delivery. Integrating technologies with its new SAP system helps Dade Behring deliver the goods on time.
As new technologies enter the auto ID field, the basic premise of "automation" is in danger of getting lost in the maze of enticing new technologies. Although technology fanatics jump to use these new products, the mainstream buyers shy away from complexities that require too much of a change in how they do business. Most successful technologies have the complexities buried on the inside and have a simple interface on the outside. Consider that the Internet didn't expand to the masses until the World Wide Web interface was delivered to the average consumer via browser software. It removed the complex set of text-based commands to the inside, presenting a simple and uncluttered interface to the outside.
Simple Interfaced Needed
Bar code is the opposite: providing simplicity on the inside and presenting an unintelligible pattern of bars and spaces on the exterior. Conversely, OCR moves the complexity to the inside of the scanner and leaves the outside, end user interface as simple as can be: "0123456789."
OCR got its commercial start during the 1950s when Standard Oil used its computers to process credit card bill payments by reading printed type. Two decades later, the first handheld OCR wands were in JC Penney's stores nationwide, enabling POS and inventory scanning. The emphasis was on simple imprints that could be read and understood by anyone.
The Emergence Of Bar Codes
A fork in the road appeared in the mid-1970s with the grocery industry's standardization of UPC symbology. Since then, bar code has become the dominant auto ID methodology, eclipsing OCR. Because bar code is a relatively simple task for a computer, it met with early success while OCR remained much more technology-intensive. Bar code data capture is also simpler because it uses a one-dimensional image whereas OCR needs a two-dimensional image. A further differentiation is that bar code can be scanned from greater distances due to its use of laser light. However, implementation of bar codes comes at a higher cost: there is often a requirement to purchase special high-resolution printers; product labels carry markings that are not human readable; and bar code requires more ink or toner. As a result, bar code users often pay more for the outward sophistication of their auto ID choice.
Wedge Simplifies Bar Code Scanning
And yet, simplicity remains a desirable auto ID goal. One of the technical breakthroughs that accelerated the adoption of auto ID equipment was the keyboard wedge. With a wedge interface, you can plug a scanner into any existing application and benefit from automation in minutes. Before the wedge, a talented programmer would have to modify the host computer to receive scanned data through a serial port. The complexity of a wedge interface is hidden; the wedge's attraction is its simple plug-and-play feature. Like OCR, the wedge provides instant automation without requiring technical involvement because users are not required to develop custom software interfaces.
Today, due to the frenetic progress of semiconductor research and development, 32-bit microprocessors combining several megabytes of memory are now affordable and available in embedded devices such as hand held scanners. Modern OCR systems scan typical print at 200 characters per second, achieving accuracy rates comparable to bar code-fewer than one in 1,000,000 characters misread. OCR can be printed at much lower resolution and typically requires one-sixth the space of a linear bar code. For example, a 15-digit bar code at medium resolution (13 mil) uses 1/2-inch of vertical space and 3-5/8 inches horizontal. OCR holds the same information in this space: 012345678901234.
Promising New Direction For OCR
One of the most promising new directions within the OCR industry is the integration with hand held digital cameras or 2D imagers. Due to the camera's increasingly popularity within the auto ID industry, OCR has now been extended to non-contact portable scanners. OCR is now even better suited to applications where bar code was impractical, too expensive or where labeling included only text or data. OCR-enabled hand held scanners are finding diverse applications in markets such as the livestock industry.
Comparing OCR To Bar Coding
This industry has been forced to address issues such as bloodline management in the wake of high profile afflictions such as "mad cow" disease. Livestock tagging requires: a robust plastic tag that must be disposable, low cost, and enough durability to withstand inclement weather. Due to low printing resolution, bar code is an impractical solution.
A better solution is to simply use OCR to read existing cattle tags. The ability to link cattle identity numbers quickly to herd management software using OCR is one of the reasons why one of the largest agribusiness VARs is closely reviewing its application. Expect to see ranch hands with OCR-enabled imagers (scanners) during the spring of 1998.
One of the imager's (scanner's) primary applications, 2D bar codes, could well serve as a perfect example of technology overkill. The adoption of this new bar code format has been disappointingly slow because potential users remain unconvinced that they require so much information on an imprint. And, of course, there's no way to read the information without a special scanner.
Bar Code Or OCR Current Applications?
The United States Post Office (USPS) is currently seeking a method to reduce the fraud in commercial metered mail markings. The USPS tendered a proposal to the vendor and user community to use a large PDF417 bar code in place of the current postmark. The community responded vociferously, citing technical barriers, such as difficulties in printing and reading and has seriously questioned the need for so much capacity for information on an envelope. I am convinced that one of the strongest arguments against the big bar code on your first class mailing piece is that it is plain old ugly. It is nice to see the post office updating, but it seems strange that it would be done by something as visually displeasing and overwrought as PDF417.
The country's largest postage meter manufacturer is already utilizing a simple anti-fraudulent OCR-based machine-readable postmark that looks like a traditional postmark. The distinction is that it houses technology on the inside — a sophisticated device that generates a unique numerical sequence is inside the postage meter and hand held OCR readers are available to audit them.
OCR & Bar Codes Can Co-Exist
The ISO committee on passport standardization is presently considering the interest among various countries to put alternate technologies, ranging from RF tags to 2-D bar codes, on their passports. In conflict with these technologies is the need among passport holders to understand what personal information is on their identity cards. If you look at your passport, you'll see two lines of human and machine readable OCR-B characters, a fact that should remain unchanged for a long time.
Clearly, bar code is an ideal choice for many applications and successfully addresses the need for a nearly indestructible non-contact automatic identification method. But, bar codes are not an infallible choice for all auto ID applications. OCR, with its technology often veiled by an attractive plastic housing, delivers far more than you ever likely thought possible. The modern ease of scanning simple printed characters suggest that perhaps it is time for everyone to revisit OCR as an auto ID technology.
Andrew Allansmith is the marketing manager of the Automated Data Entry Group at Caere Corporation.