Barcodes are widely used across a whole range of applications to uniquely identify products, places and things. This means that it is possible track and identify products through a supply chain at either a stock keeping unit level (SKU) or at a single item level. Standardized barcode symbology ensures reliable information exchange via diverse reading devices, including hand-held readers, camera systems, and mobile phones. These reading devices rely on optical recognition to decode the black and white pixels which make up a 2D barcode, or the black and white lines which make up a 1D barcode.
In many applications, invisible barcodes are used to provide unobtrusive identification and authentication such as in product track and trace. Invisible barcodes can be printed on a wide range of substrates and products with careful selection of inks and inkjet technology.
However, reading invisible barcodes presents more challenges than traditional barcode scanning. Factors like lighting conditions, barcode quality or environmental interference from specular reflections from shiny surfaces can influence the reliability of barcode reading. These challenges can result in inefficient or unreliable reading especially at high decoding speeds required for capture of codes on process lines. Conventional error correction algorithms and vision enhancement functions as part of the decoding process, often are not successful in resolving the more challenging barcode reading tasks.
Artificial intelligence (AI) and machine learning (ML) can offer solutions that can help to overcome these challenges. AI is comprised of algorithms that can sense, reason and adapt depending on the data. ML is a subset of AI where the performance of the algorithms improves as they are exposed to more data over time.
The use of AI in barcode scanning enables improvements in image recognition. AI enhances image recognition in barcode scanning, significantly improving the ability to read damaged or incomplete barcodes. Traditional scanners often struggle with blurred or partially visible barcodes, but AI-powered systems can analyze patterns and intelligently predict missing or damaged portions, filling in the gaps.
Machine learning provides barcode scanning systems with the ability to ‘learn’ from data and make decisions based on that learning, for example to compensate for differences in background print.
This paper will discuss the transformative role of AI and ML in barcode scanning for security marking. It will focus on the detection of invisible fluorescent barcodes, and how these new approaches to barcode scanning have enhanced reliability and overcome traditional limitations