CHAPTER 1
Introduction
Bar code is a technology that ubiquitous in modern life. Widely implemented in theretail marketplace, it has also been adopted in a broad range of diverse non-retailapplications. Bar code and the equipment to print and read it is increasingly being treatedas a "commodity", but there is still significant technical innovation occurring in the field.
The term "bar code" has come to include many different forms of contrasting opticalmarks that encode information, but we will start out by looking at the traditional definitionof a bar code symbol. A conventional bar code symbol consists of a series of varying widthparallel, adjacent bars and spaces. Predetermined width patterns are used to representactual data in the symbol. A simple bar code symbol is illustrated in Figure 1-1.
Bar code can be thought of as a printed version of the Morse code, with narrow barsrepresenting dots, and wide bars representing dashes. To read the information contained ina bar code symbol, a scanning device – such as a light pen – is moved across the symbolfrom one side to the other, as shown in Figure 1-2. As the scanning device is moved acrossthe symbol, the width pattern of the bars and spaces is analyzed by the reading equipmentand the original data is recovered. Some scanning devices do not require the operator toprovide the scanning motion: an electronic scanning system or arrangement of movingoptical elements allow the symbol's bars and spaces to be sequentially examinedautomatically. More advanced equipment uses machine vision technology to "take apicture" of a symbol and decode the information directly.
Although this book's title refers to "bar code," we will broaden the scope to include otherforms of optical symbols that are specifically designed for reading by machine means.
Bar code is an automatic identification technology. It allows real-time data to becollected accurately and rapidly. Modern 2-D bar code symbologies have the ability tocontain a large amount of data that can rapidly be extracted. But bar code by itself does notsolve problems. The combination of bar code with appropriate computer hardware andapplication software creates the potential for improving performance, productivity, andultimately, profitability.
This book provides a thorough background of the technology of bar coding. It reviewsapplicable equipment and outlines selection criteria. It also presents information ontechniques and approaches to system integration and examines several applications in abroad range of industries.
CHAPTER 2
Data EntryTechniques
Computers have become an integral part of almost all business operations. They areactively used for planning, controlling, producing, tracking, and analyzing most aspects ofcommerce. The ever-decreasing cost and size of computers has allowed them to penetrate awide variety of businesses, institutions, agencies, organizations, and homes.
A piece of computer hardware by itself is not exceptionally useful. In order to beproductive, a computer must be equipped with software suitable to the particularapplication. The effectiveness of the computer hardware/software system is a function ofthe input data that is provided to it. In order to maximize the benefit from a computer,timely (ideally, real-time) and accurate data is required.
There are many data collection techniques that can be used to provide data to computers,and some of these will be reviewed in this chapter.
2.1 Manual Methods
The traditional method of entering data into a computer system has involved manuallykeying in information (using a keyboard) that has previously been gathered on sheets ofpaper. Studies show that the error rate with this technique is approximately 1 error forevery 300 characters entered. Obviously, every data transaction requires that a humanoperator be involved, so there are cost, reliability, and availability issues to be considered.
Manual keying does not provide real-time data entry, since the data being enteredusually reflects events that occurred in the past. Because data is often gathered first onpaper forms and then transcribed via keyboard, several opportunities for making data errorsexist.
2.2 Automatic Methods
To offset the disadvantages of manual entry methods, several automatic data entrytechnologies have been developed. In this usage, "automatic" refers to the fact that a singleentry event can result in the capture of a stream of data (from a single character to dozens ofcharacters). In this definition of automatic data capture, a human operator may or may notbe a part of the actual entry data activity.
In evaluating automatic identification techniques, there are two important parameterswhich must be considered. The first is substitution error rate (SER), often referred to as the"error rate." This term describes the probability that a given scanned character contains anerror. The number of errors to be expected in a given application is equal to the substitutionerror rate times the number of characters scanned.
The second parameter is the first read rate (FRR). This term (expressed as a percentage)refers to the probability that an attempt to capture data (a "scanning attempt") will result indata being captured on the first attempt. As an example, if 1000 symbols are to be scannedusing a system with a 75 percent first read rate, then approximately 1333 scanning attemptswill be required to capture the data from all of the symbols. For a given technology, there isusually a strong inverse relationship between first read rate and the substitution error rate.
Other questions to consider when reviewing different data entry technologies include thefollowing:
1. Is the data "printed," or is it recorded in some other fashion? Information can bemachine read using several different approaches: optical, mechanical, electrical, ormagnetic. Printed data is simple and economical to use, since the same printingprocess can also be used to create human readable text and graphics.
2. Can the object be "scanned" at a distance or in the presence of motion?
3. Does the automatic identification technology permit modification of the recordeddata? The ability to record as well as read information offers the greatest flexibility,but may introduce concerns about data integrity.
We will now review many of the automatic identification technologies that have beendeveloped over the years.
2.2.1 Punched Cards
Patented by Henry Hollerith in 1887, punched cards are one of the oldest forms ofautomatic identification. A stiff piece of paper is punched with holes in specific locations toencode data. Keypunching data into a card is obviously a manual process, but the ability tobatch read a stack of cards into a data processing system confirms that this is an automaticidentification technology.
Punched cards were originally used for programming mechanical textile looms, butwhen early electronic computers were developed, their use continued as a data entry device.Punched card use has declined rapidly since the early 1970s.
2.2.2 Mark Sense
Mark sense involves the presence or absence of printed marks in fixed locations on adocument. The marks are equivalent to the punched holes in tabulating cards. Theprinciple of mark sense has been broadly applied to voting machines and the automaticscoring of students' tests.
Mark sense was successfully used in material handling and warehouse sortation prior tothe widespread acceptance of bar code.
Mark sense is a transitional technology between punched cards and modern bar codes.The black marks on a mark sense document bear a one-to-one relationship with the holes ina punched card. When the cost of processing electronics declined dramatically, it becamemore practical and versatile to line up the marks in a single row, thereby forming a bar code.
2.2.3 Optical Character Recognition
Optical Character Recognition (OCR) encompasses the machine-reading of charactersand text that are also intended to be read by humans. Today, OCR is usually associatedwith reading sheets of text.
There has been considerable Research and Development over many years intotechnology that can "read" conventional printed text. A big motivation for this developmenthas been the requirement for entering large amounts of information into computers fromexisting printed documents, but one of the first major applications was for labelling andreading tags placed on merchandise in retail stores. In the early 1970's, the NationalMerchant Retailers Association (NRMA) chose OCR as its preferred technology for in-storelabelling and scanning - this type of equipment was deployed in many Sears stores at thetime. Another big application of OCR was the processing of checks and credit card salesslips. When traditional OCR was used to label products or documents in the mid-1970s, ahand-operated scanner was usually employed. Some operator skill was required, and thefirst read rate could easily be less than 50 percent. If good quality labels were used, carefuloperators who had received suitable training could achieve a first read rate of 80 percent ormore. Recent developments in imager-based hand-held scanners have offeredimprovements in this area.
It is possible to read hand-printed characters using special software and appropriatehardware, but far better results are achievable if the characters have been printed using aspecific, pre-determined font. Because of the limited performance of early OCR scanners, afont known as OCR-A was developed which was easier for machines to decipher, while stillbeing readable by humans. As the technology improved, a font known as OCR-B wasdeveloped, which more closely matched conventional printed text, making it easier forhumans to read while still having characteristics that allowed for reasonable scannerperformance. Examples of both fonts are shown in Figure 2-1.
Unlike bar code, it is not possible to extract the information contained in a printed OCRAor OCR-B message by simply taking a "single slice" through the characters. This isbecause OCR characters need to be examined over their complete area in order to decodethem. Whereas conventional bar code can be considered as a one-dimensional technology,OCR is a two-dimensional technology. OCR fonts can be printed by several differenttechniques and are designed to be readable by humans as well as machines. When used toread pages of OCR-printed text, an automatic page scanner can quickly capture all of thedata, while exhibiting an error rate of 1 character out of approximately every 10,000 scanned.
Recent developments in font-independent scanning equipment have been made possibleby advances in recognition algorithms and the decreasing cost of computing, but thistechnology has not yet had much impact on traditional automatic identification applications.The primary market for this technology has shifted: font independent OCR is nowprimarily used as an entry technique for word processing and desktop publishing systems.Typically, a page scanner is used in conjunction with an OCR software application runningon an attached computer.
One OCR application that is growing is its use in passport control. Many countries printspecific identification data in a prescribed position in passports, using the OCR-B font.Fixed-position specialized scanners are then used at border control points to collect datafrom these Machine Readable Passports (MRP).
2.2.4 Magnetic Ink
Magnetic ink character recognition (MICR) is the technology commonly employed in themarking of bank checks (see Figure 2-2).
This technology uses a highly stylized font that is printed with an ink that has magneticproperties. Although the information could be read optically, MICR characters are usuallydecoded with fully automatic magnetic scanners. As in all magnetic technology, thereading equipment makes contact with the characters to be read.
This technology is well entrenched in the banking industry, having been used for manyyears. At the time MICR was introduced for bank checks, this was the only technologyavailable for automatic entry of the account data. Because of the special ink and complexreading equipment, MICR is not used or suitable for general purpose applications outsideof banking.
2.2.5 Magnetic Stripe
It is possible to encode a great deal of information onto a magnetic stripe, similar to theone found on the back of most credit cards. Information is stored as a series of regions withdiffering magnetization, just like computer tapes or floppy disks. Data can be written to orchanged on a magnetic stripe, adding flexibility to an information system.
Magnetic stripes have not been widely adopted for general tracking applications becauseof:
• The unavailability of noncontact scanning equipment.
• Environmental considerations.
• The inability of conventional printing methods to encode magnetic information.
• Higher labeling costs compared to optical technologies that use images printedonto paper substrates.
2.2.6 Voice Recognition
Equipment is available that can "understand" spoken speech. This technology is wellsuited for certain manned applications (like baggage sortation) where operators need bothhands free. Voice recognition, however, does not fit our definition of automaticidentification: part numbers will have to be spoken character-by-character. Furthermore,commercially available equipment must be "trained" for each operator and has a ratherlimited vocabulary. The operator is an integral part of the data capture process and is themain cause of all errors. The continuing evolution of electronic processors is resulting inprogress toward the ultimate goal of speaker-independent recognition of connected speech,but there is a long way to go yet.
2.2.7 Machine Vision
Machine vision systems are used in many manufacturing companies to sort, to inspect, orto measure products automatically. They consist of a high-resolution television camera (orequivalent) interfaced to a computer via signal processing circuitry. Considerable softwarecomplexity is involved in most machine vision systems. This equipment is usually custom-tailoredfor each application.
In order to differentiate between externally physically similar objects, some form ofoptical marking is still required. This can take the form of conventional characters or a seriesof arbitrary marks. A machine vision system can perform automatic identification via barcode or OCR symbols, but its primary function is inspection and sortation.
A later chapter in this book will describe "Imager-Based Scanning," which is a specializedbar code scanning technique with many similarities to machine vision.
2.2.8 Radio Frequency Identification
Automatic identification systems using radio frequency (RF) can read data from tagsthat are not even optically visible to the system. A radio signal is transmitted toward thetag, and it responds with a radio signal that is modulated with information stored in the tag.One of the initial applications involved the tracking of livestock, but many applications intransportation and manufacturing have subsequently developed.
Tags can be preprogrammed with data, or some tags can allow the data to be changed inresponse to commands modulated onto the interrogating radio signal. Long rangeprogrammable tags require an internal battery (these are referred to as "active tags");nonprogrammable tags and shorter range programmable tags usually derive their operatingpower from the interrogating beam (these are known as "passive tags").
Most RF identification uses two different frequency ranges: either very low frequency(below 300KHz), or very high frequency (above 200 MHz). The high frequency systemsallow longer ranges and higher data rates, but the costs are also higher. Low frequencysystems allow more casual tag orientation. Some specialized systems are now emergingwhich use frequencies in the range of 13.5 MHz.
The simplest tag is one that contains only a single bit of information. This is sometimesused in security applications or as an anti-shoplifting device. In industrial applications, RFtags often can contain several thousand bits of information.
