OCR_paper OCR Whitepaper

: Microscan Ocr Whitepaper OCR_whitepaper ecc2ad65-cb86-45cd-87a1-93a70cdd4f52 _att

Open the PDF directly: View PDF .
Page Count: 5

Understanding Optical Character Recognition

Optical Character Recognition, commonly known as OCR, is distinct from linear and 2D symbologies in

that it is simultaneously machine-readable and human-readable. OCR does not replace the more robust,

secure options of linear and 2D symbologies. For traceability applications, 2D symbols offer the highest

level of data security and reliability, and linear symbols offer an intermediate level of security and reliabil-

ity. OCR alone offers the least. OCR is most effective when used to complement linear and 2D symbols.

A History of Optical Character Recognition Technology

Optical Character Recognition technology has been used extensively in commercial applications since

the 1970s. In the early 1970s, a company in Dallas, Texas, called Recognition Equipment, Inc., devel-

oped a high-speed system for reading credit card receipts from gasoline purchases. At the time of the

transaction, a receipt would be imprinted with the customer’s account number (typically embossed on

credit cards in OCR-A font). The merchant copy of the receipt would then be sent to a processing center

where equipment provided by Recognition Equipment, Inc. would read the OCR-A account numbers on

the receipts at document speeds of 45 to 55 feet per second.

By the late 1970s, OCR-B was being used on payment stubs for automated payment processing. Some

utility companies, such as Southern California Gas, still use this system. Also in the late 1970s, Recogni-

tion Equipment, Inc. released a handheld OCR reader. The product was developed in response to the

retail community’s desire to switch from punch-hole price tags to price tags with OCR strings. Shortly

after Recognition Equipment, Inc. introduced the fi rst handheld OCR reader, Robert Noyce (co-founder

of Fairchild Semiconductor and Intel) founded the Caere Corporation. Caere (later acquired by ScanSoft,

Inc.) introduced its own handheld OCR reader in 1977. As a result of the new handheld OCR technology,

several large retailers, including Sears, JCPenney, and Kmart converted to OCR-A price tags between

1980 and 1983. Sears and JCPenney alone purchased 50,000 readers. OCR-A was used on price tags

until 1987, when the retail community selected UPC as their standard.

Since the mid-1980s, OCR technology has been adopted in a wide variety of applications, including

remittance processing, passport processing, semiconductor manufacturing, automotive and aerospace

manufacturing, secure document processing (checks, fi nancial documents, bills), document handling,

postal tracking, publishing, food packaging and consumer goods packaging (batch codes, lot codes, expi-

ration dates), and clinical applications.

APPLICATION

WHITE

PAPER

1 www.microscan.com

Figure 1: Examples of OCR used in a variety of applications

APPLICATION

WHITE

PAPER

In remittance processing applications, individual payment stubs are printed with an OCR string, which is

usually decoded by a fi xed-mount reader in an automated environment. In passport processing, traveler

information is encoded in two lines of OCR text. The use of OCR on passports was introduced in 1983 as

part of an international convention. In 1984, Caere Corporation developed the fi rst passport scanner for

the U.S. State Department, and some are still in use today. The use of OCR on passports may diminish

as bio-chips and other identifi cation technologies gain greater currency in the coming decades.

Wafers and lead frames in semiconductor manufacturing applications are often marked with OCR strings.

In automotive and aerospace manufacturing, OCR strings are placed on parts and sub-assemblies as

direct part marks (laser etch, chemical etch, dot peen, etc.)

OCR vs. Bar Code Technology

OCR and bar code technology are both data capture methodologies, and each has advantages and

disadvantages. The primary advantage of OCR is that it encodes information in a format that is simultane-

ously machine-readable and human-readable, while linear and 2D symbols are only machine-readable.

Data encoded in an OCR string does not require a secondary machine-readable symbol. Data encoded in

linear and 2D symbols is considerably more reliable, however. OCR has an inherently high rate of charac-

ter substitution (particularly the OCR-A and OCR-B fonts)—not typically a concern when using linear and

2D symbols, which offer greater data integrity. Check characters are often embedded in OCR data fi elds

and then calculated by OCR readers or vision systems to avoid substitution errors in data output. Many

OCR readers have the ability to re-try the decode process a predetermined number of times, since sub-

stitution rates of as many as one of every 3,000 characters are expected in OCR applications. (See Using

Checksums to Reinforce OCR Data Integrity.)

OCR Templates vs. “Teachable” OCR Systems and Optical Character Verifi cation

There are different ways to integrate OCR into an application, and different systems for processing OCR-

encoded data. OCR templates and OCR fonts are the simplest and most reliable option. Examples of

some common OCR fonts are shown below.

OCR-A

OCR-A is a relatively reliable font that supports an alphanumeric character set, along with some additional

ASCII characters. It complies with the character shape, size, and printing position requirements for the

ANSI INCITS 17-1981 (R2002) standard, which can be purchased from the ANSI website: http://webstore.

ansi.org.

OCR-B

OCR-B is less reliable than OCR-A, but its less angular characters are generally considered to be more

aesthetically pleasing. It complies with the character shape, size, and printing position requirements for

the ANSI INCITS 49-1975 (R2002) standard, which can be purchased from the ANSI website: http://web-

store.ansi.org.

2 www.microscan.com

APPLICATION

WHITE

PAPER

MICR E-13B

MICR E-13B is used primarily in the banking industries of the U.S., Canada, Puerto Rico, the UK, and

Panama. It is most commonly seen at the bottom of personal checks, where account information is

encoded using magnetic ink (MICR is an abbreviation of “Magnetic Ink Character Recognition”). MICR

E-13B complies with the character shape, size, and printing position requirements for the ANSI X9.27-

2000 standard, which can be purchased from the ANSI website: http://webstore.ansi.org.

MICR CMC-7

MICR CMC-7 is used primarily in the banking industries of France, Spain, and most Latin American coun-

tries. It is most commonly seen at the bottom of personal checks, where account information is encoded

using magnetic ink. MICR CMC-7 complies with the character shape, size, and printing position require-

ments for the ISO 1004:1995 standard, which can be purchased from the ISO website: http://www.iso.org.

SEMI M12

SEMI (Semiconductor Equipment and Materials International) is used for wafer and lead frame mark-

ing in the semiconductor manufacturing industry. It complies with the character shape, size, and printing

position requirements for the SEMI M12-0706 standard, which can be purchased from the SEMI website:

http://www.semi.org.

OCR Templates

OCR templates defi ne several parameters, including the OCR font that is used, layout of OCR text (in a

row, in a column, etc.), the number of characters in a row of OCR text, the total number of rows, and the

total number of characters in all the rows. Each character position in a row is specifi ed as an ASCII value,

a group of ASCII values, a wildcard character, or a combination of known ASCII values and wildcard char-

acters. Limiting the variables of character type and character position as much as possible improves the

reliability and effi ciency of OCR applications.

Software-Confi gurable OCR Parameters

The task of optimizing OCR reliability and effi ciency is vastly simplifi ed by the current generation of con-

fi guration software. Many current providers of OCR technology, such as Microscan Systems, Inc., offer

intuitive software interfaces to assist users in setting up OCR applications. See Figures 2 and 3 on the

following page.

Teachable/Trainable OCR Systems and Optical Character Verifi cation (OCV)

In contrast to the relative simplicity and reliability of OCR templates, “teachable” or “trainable” OCR sys-

tems are a technologically impressive but potentially unreliable option for OCR applications. Typically a

feature of higher-end machine vision, teachable OCR systems can be trained to recognize characters in

any user-defi ned font—not just fonts that are created specifi cally for optical character recognition (OCR-A,

OCR-B, MICR, SEMI). OCR systems can be taught to recognize a full character set in any font created for

any language. The disadvantages of this type of OCR system are the labor-intensive integration process

and the decrease in reliability when using fonts that are not created specifi cally for OCR applications.

3 www.microscan.com

APPLICATION

WHITE

PAPER

4 www.microscan.com

Figure 2: Example of OCR template interface

Figure 3: Example of OCR confi guration functions

OCR is enabled or disabled using the

OCR Enable check box.

Additional check boxes allow the user

to refi ne OCR confi guration so that the

reader will look only for the selected OCR

attributes.

The Orientation dropdown menu allows

the user to select the direction of OCR

strings (Up, Down, Left-to-Right, Right-

to-Left).

The Templates dropdown menu allows

the user to select a preset OCR tem-

plate (Passport, ISBN, Price Field, MICR

E-13B). The Row Count dropdown menu

allows the user to set the number of rows

in the OCR string (1 - 3).

If “User-Defi ned” is selected on the

Templates dropdown menu, the fi elds

shown at left can be used to refi ne OCR

parameters even further. The Font menu

allows the user to select the OCR font

used in the application (OCR-A, OCR-B,

Both OCR-A and OCR-B, MICR E-13B).

The Character Count menu sets the

number of characters (1-20) in the OCR

string. Individual positions in the OCR

string can then be defi ned using the text

boxes shown at left.

The buttons shown at left allow the soft-

ware to Receive settings from the reader,

to Send settings to the reader without

saving, or to Send and Save settings.

The confi guration functions available in

the OCR interface are also listed in a tree

control, which allows users to enable or

disable individual parameters.

APPLICATION

WHITE

PAPER

Optical Character Verifi cation (OCV) is one way to address the problem of reliability in teachable OCR

systems. Once the desired specifi cations have been taught to an OCR reader, OCV software can verify

that characters are printed to match the user-defi ned specifi cations, can ensure that data is encoded cor-

rectly, and can guarantee that labels are placed in the correct orientations on the correct items.

Using Checksums to Reinforce OCR Data Integrity

The purpose of using checksums with OCR is to reduce the likelihood of character substitution errors.

OCR systems commonly provide users with various checksum options. Checksum types (row or block),

weight schemes, and modulo values are all ways of ensuring the correctness of data encoded in OCR

strings.

Conclusion

Although OCR was originally developed decades ago, it continues to

be used in a broad range of application environments, and continues

to be supported by a wide variety of products and systems—from

high-end machine vision to more compact, easier-to-integrate solu-

tions such as Microscan’s Quadrus® MINI series imagers.

Some OCR applications, particularly those using templates, can be

fully supported by a lower cost OCR solution such as the ones pro-

vided by Microscan. Other applications may have a greater number

of variables, requiring a more complex vision system. Thorough

evaluation of all the attributes of the target application is necessary

before choosing an OCR solution.

5 www.microscan.com

Microscan Systems, Inc.

Tel 425 226 5700 / 800 251 7711

Fax 425 226 8250

Microscan Europe

Tel 31 172 423360 / Fax 31 172 423366

Microscan Asia Pacifi c R.O.

Tel 65 6846 1214 / Fax 65 6846 4641

www.microscan.com

E-mail: info@microscan.com

Tech Support: helpdesk@microscan.com

Copyright ©2008 Microscan Systems, Inc. All rights reserved.

ISO 9001:2000

Certifi ed QMS