Optical Character Recognition (OCR) is known for its ability to read all character types - machine print, handprint and cursive - in many style variations. The technology automates the reading of addresses on letters, flats and parcels. It was developed to specifically address the needs of the postal market and has taken automatic sorting to levels unheard of just a few years ago.

High-volume mailers, postal operators, fulfillment service bureaus and digital mail center automation vendors streamline and automate the sortation of the full mail stream - including letters, flats, parcels, forwarded mail, business reply mail and bundles with solutions developed and implemented for outgoing, incoming and interoffice mail sorting.

Next generation universal address recognition technology for the postal market incorporates the latest achievements in artificial intelligence and uses the combination of multiple independent engines and contextual information to read machine print and handwritten addresses with unprecedented performance in deciphering difficult, poor quality and hard-to-read input material typically rejected by conventional OCR systems. This next generation technology helps posts to address challenges faced in gaining further advancements in mail sorting and in obtaining new business opportunities. The technology works in mail-forwarding and sorting applications pertaining to parcels, bundles, flats and letters and can be easily adapted to country-specific address recognition requirements. In particular, the universal OCR software has been used to develop handwritten and machine-print address recognition software for a number of countries including Canada, France, Germany, Italy, Netherlands, Russia, Sweden, UK and United States and has accomplished each project within a short period of time and with relatively low expense.

Today, universal OCR technology has become a requirement for postal operators that want to take their systems to the next performance level.

Universal OCR
Universal OCR serves as a backbone for a wide range of postal and mail sorting applications. It is not only used to decode the whole variety of domestic addresses up to the highest depth of sort, but also to solve such challenging problems as mail redirection, processing of incomplete addresses and reading foreign mail. All of these applications go beyond mere address recognition to perform a broader range of tasks, including more sophisticated data analysis. In performing mail redirection, the technology reads and matches personal/business addressee names contained in the destination address block with names contained in the USPS Change of Address (COA) database, locates and reads the return address, executes Address Change Service (ACS) and endorsement determination.

For instance, processing mailpieces with incomplete addresses is part of a bigger undeliverable-as-addressed (UAA) problem that causes significant losses for USPS and other postal operators throughout the world and is addressed by universal OCR technology. Such mailpieces typically require additional handling: if an address cannot be automatically resolved, the mailpiece is sent for manual processing, often times to conclude that the address is unreadable, and that it cannot be processed further. The latest OCR technologies include an arsenal of sophisticated techniques enabling a conclusive judgment about the quality of address information (for example, technology can determine that an address is incomplete or inaccurate), thus eliminating the need to send a mailpiece for manual keying.

Other potential uses for the universal OCR technology include:
· Detection of overlapping mail pieces (so called "double feed"), even where the shift is very small, to ensure that the covered mailpieces are not lost or sent to a wrong destination.
· Detecting and recognizing "Return to sender" stamps, which may be randomly located on mailpieces and are often of a very poor quality.
· Processing foreign mail, discrimination between domestic and foreign mail
· Processing addresses written in foreign languages

Inbound and Outbound Mail Sorting for German Mail
Universal OCR technology is used for inbound and outbound international mail sorting in Germany. This application actively exploits and benefits from a broad range of universal OCR features and capabilities. Outbound mail stream processing is limited to reading the country name on mail leaving Germany. Therefore, OCR locates and reads a country name, which may be written in German as well as in the language of the destination country. For example, a letter addressed to the United States may have the country name written as "USA" or "Vereinigte Staaten." A letter addressed to the Ivory Coast may have "Côte d'Ivoire" or "Elfenbeinküste." It is important that a postal sorting machine be able to correctly interpret any variation, in writing, of the country name.

Another challenge is to reliably discriminate between domestic and foreign mail. Typically, foreign mailpieces are processed separately and are sent abroad. If a domestic address is interpreted as a foreign address, the mailpiece is sent abroad and will either be lost or, at best, reach the addressee with a significant delay. Universal OCR meets the most strict error rate requirements to differentiate between domestic and foreign mail.

Inbound mail processing implies sorting German addresses for mail coming from different countries. In this case, universal OCR uses standard Deutsche Post coding rules ensuring the required finalization levels: "Outward" sortation levels include recognition of a five digit German postal code as well as city name on a mailpiece. Optionally, if database content is available, the depth of automatic coding may be increased from an outward level to an inward level that decodes addresses to an eight-digit German postal code (street name recognition) and even to a "House Number" level, decoding all mailpieces coded to an 11-digit German postal code. In the latter case the house number is recognized on a mailpiece.

Implementation of the Universal OCR
Implementation of the universal OCR for commercial mail sorting in Germany is another example of benefits that advanced technology brought to a leading German mailer, processing over 600,000 items per day in an eight hour time window. Of these 600,000 pieces, 250,000 are processed using the new universal OCR. The automated solution efficiently processes a mixed mail stream of letters and flats providing significantly increased performance in machine-printed and handwritten address reading compared to a previously installed system, and allowing private post mailers to capitalize on maximum efficiency of mail processing. In particular, the system enabled significant reduction of first pass rejects providing over 30 percent savings per item compared to video encoding. Using the latest address recognition and interpretation technology, it is possible to achieve savings even on the most illegibly printed and handwritten addressed mail pieces.

Due to the higher depth of coding, the solution also enabled walk sequence sorting into smaller merged blocks, street snippets (instead to individual households) in order to cut processing time. As a result of the integrated OCR solution, the postal service provider has increased first pass finalization and has drastically reduced second pass video coding efforts, saving time and money. The solution allows customers to automatically read over 94% of addresses on mailpieces. This includes 15% of handwritten addresses, resulting in over 30% savings per mail item over manual data entry (video coding).

Postal services and mail processors continue to be faced with the same challenge every day - getting millions of mailpieces delivered on time. On top of that, they are also being confronted by increased competition in the form of newer, faster and higher quality services offered by others today. Accurately reading information contained on mail pieces is necessary to provide successful mail processing and innovative services to customers. Using advanced technology to ensure the most complete and detailed understanding of the content of all data, printed or written, on a mailpiece is now a key element to address these challenges and gain a competitive advantage.

Kaz Jaszczak, with more than 25 years experience in research and product management, is Director of Product Planning and Operations at Parascript, LLC., an address recognition and interpretation company. He can be reached at kaz.jaszczak@parascript.com or 303-381-3153.