Recovering considerable losses through unpaid revenue

Loss of revenue through a combination of insufficient postage or counterfeit stamps is a familiar problem throughout the postal industry. Suffering ongoing frustrations with this problem, our client approached us to see whether we could help alleviate this.

Client situation

Loss of revenue through a combination of insufficient postage or counterfeit stamps is a familiar problem throughout the postal industry. Suffering ongoing frustrations with this problem, our client, an international postal and logistics organization, approached us to see whether Prime Vision could help alleviate this.

Increasingly, letters and parcels are under-franked when the sender uses stamps to pay for the shipment. Additionally, mailed items are often franked with counterfeit stamps or stamps that were already taken out of circulation. Others are sent, particularly internationally, where the paid value of the stamp did not match the required cost for the destination.

Systems were in place within the customer’s operations to attempt to combat these issues, but still approximately 1% of sent items franked with stamps carried insufficient value or, even worse, none at all. This seemingly small percentage cost major international postal and logistics companies tens, sometimes hundreds of millions of Euros per year. In an industry with tight margins, these percentages could mean the difference between profitability and loss.

Challenge: locating, identifying and comparing the stamp

The task at hand may sound relatively simple: locate the stamp, identify the stamp, assess its properties and see if it matches the weight, dimensions and destination of the sent item. In reality, though, this presents a series of unique and complex problems, each requiring a mix of new techniques and technologies to be solved.
Firstly, the stamp needs to be located.
But how does a machine know what a stamp is and not, for example, a part of the address label or the packaging itself? In addition, a person can assume a stamp is pasted on the top right corner, but how does your camera system know the difference between top and bottom? The wildly varying addressing habits and tumbling journey through a mail sorting system require a solution that can adapt to the individual circumstances of a mail piece.

Once located, the stamp needs to be positively identified. Our customer had a database of almost 1000 stamps currently in circulation. Any one of them could be a match. While the design and values can vary considerably, stamps can very much look like one another, with maybe only a slight color variation or a gradually updated silhouette of an ageing monarch. A well-loved, long-standing design may remain in place, but the printed value may increase over the years .

Next, the status of the stamp needs to be determined. Is the stamp current or has it been cancelled ? Lastly and probably most importantly, how does the value and status of the stamp correlate to the sent item’s characteristics? Was our customer being paid the correct amount, if any? All these questions need to be answered in a fraction of a second.

How we solved it

The R&D team at Prime Vision began by breaking down the solution into four separate but integrated processes. Location recognition, interrogation, and relevance.

Location is achieved through the existing technology of phosphorus detection. The solution can quickly locate a pre-impregnated stamp.

To begin the recognition process, the system instantly makes an image comparison to either immediately identify the stamp, or set it aside for further scrutiny, should any anomalies be flagged. If further investigation is needed, a more in-depth character recognition process starts, to differentiate between similar stamps, identify counterfeit discrepancies or cancelled marks.

Once the stamp is recognized and identified as valid, the face value is then compared to the weight of the item (supplied by the host sorting system) to make sure correct postage is paid.

Ensuring the solution is always relevant, the stamp image database can be quickly updated via a client configuration interface, effectively re-training the system when new stamps are introduced.

Leading the development was Senior Researcher Sjaak Koomen, “We are particularly proud of our innovation leaps in the character recognition process. Initial development led to an off-set between speed and capability. The system was super-fast, but not accurate enough , so we had to slow it down. Although the result was a great improvement on previously available solutions and the client would have been happy with the results, we thought we could do much better. In the end, we set about re-thinking the process”.

The R&D team took their inspiration eventually from forensic fingerprinting techniques. It quickly became apparent that by identifying stable reference points for each stamp, each image could be broken down to a series of unique patches of 40 pixels.

Sjaak explains. “By creating a unique fingerprint for each stamp ,only a tiny portion of the image needed to be processed. Less processing, more speed! Also, we only needed partial images for each stamp to achieve a positive result. This innovation increased the speed by up to 20%, while still allowing deep searches . Positive results leapt from 95% to 99%”.

Benefits of Indicia Vision

In an industry with narrow margins, it’s these small percentages that make the difference for profitability. The business case for the Prime Vision Stamp Recognition system is simple. The solution can save costly, unnecessary person-hours and help recover considerable losses in unpaid revenue.

A smaller national post or logistics company usually processes in excess of 10 million letters and parcels a day. Based on this number, the step-up in performance from 95 to 99% from the old to the new situation may represent an additional €2.6m per year in prevented lost revenue, and a further saving of €350,000 per year in previously wasted working hours*.

Interested in our latest news?

Subscribe to our newsletter

This field is for validation purposes and should be left unchanged.

More news