PrimeVision for Error Prevention in Barcode Scanning

From supermarkets to warehouses, barcodes are ubiquitous for identifying and tracking items. The widespread use of these “barcodes” is due to their almost absolute reliability, with read rates of over 99.9% under ideal conditions. However, during package sorting, external factors can disrupt this nearly seamless process, meaning the barcode is not read, and consequently, packages are rejected. To minimize such incidents and reduce the number of rejected packages, Prime Vision has developed the unique solution Barcode Vision.

By: Sjaak Koomen, Senior Researcher at Prime Vision, a global leader in computer vision integration and robotics for logistics and e-commerce, reports.

(Delft) In warehouses and fulfillment centers, barcodes are used to identify packages and sort them by destination. Typically, an arrangement of five cameras positioned around a conveyor belt scans the barcode. The package is then directed to the appropriate area. This works for most objects, but a small number of barcodes are not read on the first attempt.

There are various reasons for this: The barcode may be so damaged that it can no longer be recognized. The surface of the package may be uneven, such as a deformable plastic bag, a cylindrical object, or a subsequently applied lid with a label. The barcode may only be partially visible if the label has been stuck around a corner. Printer errors or an empty ink cartridge, as well as camera issues, can also be responsible for a label being unreadable.

Sorting Issues with Rejected Packages

In all the mentioned cases, the packages cannot be sorted, requiring intervention. The first option is to scan the barcode again and send the goods through the warehouse once more, which reduces sorting capacity. If this fails, the shipment is rejected. A warehouse employee must rescan it or enter the ID number, print and affix a new label, store the package, and reintroduce it into the sorting system. The procedure is essentially the same as a barcode reading error in a supermarket and costs time and money.

A practical example: Suppose in a large sorting center that processes 40,000 packages per hour, 5% are rejected packages. This corresponds to a reduction in capacity to 38,000 packages per hour, but it also means that these 5% must be handled manually. In the described case, 2,000 rejected packages must be sorted manually per hour. With a processing time of 20 seconds each, 11 employees are needed to handle all returns^[1]. These capacity losses and the additional labor costs add up to millions in lost revenue year after year. Fewer barcode reading errors are therefore highly desirable, especially for large sorting operations.

Decoding Unreadable Barcodes

Barcode Vision is a proven digital solution for reducing reading errors. Here, an additional, specialized barcode scan provides a second chance to identify the item before it is sent back for manual processing. Unlike conventional systems, Barcode Vision employs multiple methods to reconstruct damaged or unreadable barcodes.

By analyzing the barcodes used at the respective location, the system can learn which types of barcodes to expect. If a symbol is missing at the beginning or end, it can be reconstructed and validated. Additionally, Barcode Vision can read text under the barcode and combine it with the symbols to ensure successful recognition.

For accurate reconstruction, Barcode Vision not only relies on customer information but also utilizes deep learning technology. The system, programmed in-house by Prime Vision, can intelligently reconstruct barcodes from blurry images, thus overcoming camera or printing errors.

A Whole Package of Remedies

Integrating Barcode Vision is straightforward. The system is compatible with existing scanning cameras, allowing the issue of unreadable barcodes to be resolved without replacing or retrofitting hardware or conveyor technology components. Barcode Vision has an interface for a smart integration management system (SIM) that captures images and results from existing barcode readers. If the standard reader fails, this information is sent to Barcode Vision to initiate a second attempt. The system not only has the widespread Prime Vision interface but is also provided as a Windows Dynamic Link Library (.dll) for comprehensive compatibility.

The success rate of the system can be further improved by identifying specific customer issues. Eliminating printing defects, optimizing packaging for the reader, and a larger scanner area help the software minimize the number of failed reading attempts.

The Second Chance for Unreadable Barcodes

Barcode Vision is not a cure-all for unrecognized packages, but it is certainly advantageous in every case. The success of the technology also depends on the sorting operation, but even a 25% reduction in returns means significant cost savings each year due to less manual intervention, higher storage capacity, and more efficient processes. With this unique specialized solution that reads “between the lines” for barcodes, sorting operations now have the opportunity to significantly reduce their return rates.

[1] 40000 packages/hour * 0.05 rejection rate * 20 seconds ÷ 3600 seconds/hour = 11.111

Photo: © PrimeVision/Adobe Stock