For many years, image processing and address recognition technologies have been used to speed up mail sorting and reduce manual data entry costs. The need for automation has only increased with today's competitive postal business, changing customer behaviors, and uncertain mail volumes. Additionally, continued growth in parcel volume is sparking greater need for automatic parcel sorting. Since parcel processing is labor intensive, reducing costs associated with parcel sorting and delivery has also received increased urgency.

Postal and courier parcel sorting centers currently introducing or upgrading parcel sorting systems are at an advantage over mail centers that automated letters and flats in past years. Today, more mature technology and business models provide higher savings to make solutions viable. Furthermore, accumulated experience can help in arriving at the most efficient and cost-effective solutions for automated parcel sorting.

In particular, innovative universal Optical Character Recognition (OCR) solutions, already implemented in letters and flats sorting automation, may be adapted and successfully applied to the shipping of parcels. These modularized systems are not tied to specific hardware platforms and can be easily implemented to provide an accurate, fast, flexible, and cost-effective solution in an existing environment. Until recently, suppliers were offered only integrated solutions that included both OCR and transport components. This tight integration of sorting equipment and OCR modules hampered the periodic replacement or update of just one of the components. An OCR/Video Coding Systems (OCR/VCS) open interface standard was created over the past two years to enable postal operators to work with different suppliers for needed replacements or expansions to sub-systems without incurring significant engineering costs. Universal OCR systems that may be considered separately from the upgrade of sorting equipment make it possible to unify automation efforts and provide address recognition improvements for several mail stream types simultaneously.

A Host of Benefits
There are other advantages of universal technology that are worth considering. First, the capability of an OCR engine to provide high performance levels on both machine-printed and handwritten addresses makes automation investments more viable despite the ratio of handwritten and machine-print addresses in the full mail stream. Second, universal technology automatically identifies and locates address blocks and labels on any parcel despite its format. It also requires comparatively lower expenses to adapt (for address formats and structure, coding rules, unique handwriting styles, etc.) for specific countries. This approach requires lower development expenses and makes even small business opportunities feasible and attractive.

Universal OCR solutions were designed to answer the entire range of market needs and proved their efficiency in a number of postal and parcel automation projects worldwide. The software incorporates the latest achievements in artificial intelligence in areas such as neural network technology, fuzzy logic, hidden Markov models, multiple independent engines, and contextual information. Using these innovative methods, universal OCR successfully resolves address location and recognition challenges that parcel images pose. In particular they allow the software to successfully solve the task of "Region of Interest" (ROI) location. Though often underestimated or taken for granted, ROI location is a necessary step in applications in postal and courier parcel sorting and is key to the success of the solution overall.

In cases where parcel images contain several blocks of information, like sender's address, company logos, and barcodes, as well as the destination address, ROI location is one of the main challenges. Finding the small address block amidst different pieces of data should be approached in a very precise way, yet the process must have flexibility to accommodate different applications. First, it is necessary to separate the image of a mailpiece from the background. Here, an ROI is the image of the mail piece, which is found within a larger image, which also captures a piece of a conveyer belt used to transport the mailpiece.

Despite the difficulty of the task, the best modern technologies successfully locate ROI on parcels and bundles. The most advanced algorithms rely on two main approaches ROI location software:

1. Heuristic algorithms, in which a set of rules is created to describe the possible layouts of various elements of a mail piece. During the location process, an algorithm tries to classify an input layout by matching it against a set of fixed layout descriptions.

2. Neural net technologies, used to recognize relationships between items on a mailpiece. The neural network is trained on a large image set of mail pieces and uses accumulated knowledge to locate the ROI during the mail handling process.

The software separates the ROI, cleaning up the image for processing. Then, the second-level ROI represents the required information found in either a "simple" address block or a label containing an address block. Finally, for labels, which contain additional information such as return address, phone number, account number, or barcode, the destination address block is located. While processing labels, a new ROI location task is assigned to extract the destination address data only. Sophisticated ROI algorithms filter out irrelevant objects at this stage.

Universal OCR removes background noise from the parcel images and identifies address blocks and labels, as well as detects images missing a destination address. Unique label-type independent techniques can be used to locate address blocks on any parcel and provide a universal solution for a mixed stream of packages.

When the address is located, the software reads machine printed and handwritten addresses on parcel images that are often of poor quality, rotated at random angles, skewed, and often include multiple types of "noise."

Universal address recognition solutions are now available to parcel processors to locate and recognize country-specific addresses on any parcel. These solutions enable global automation and raise quality and data accuracy standards regardless of geographic location.

Kaz Jaszczak, Director of Business Development, Government Postal and Commercial Mail for Parascript, has more than 25 years experience in research and product management. He can be reached at or 303.381.3153.