Sometimes, when we attempt to scan a QR code using a high-quality smartphone camera, the process can fail. This often occurs due to the QR code being of low image quality or printed on uneven surfaces. These surfaces might include the wrappings of delivery packages or food trays that are not flat or have irregular patterns. A team from the University of Barcelona and the Universitat Oberta de Catalunya has now developed a method to make it easier to recognize QR codes in these complex physical settings.
Sometimes, when we attempt to scan a QR code using a high-quality smartphone camera, the process can fail. This often occurs due to the QR code being of low image quality or printed on uneven surfaces. These surfaces might include the wrappings of delivery packages or food trays that are not flat or have irregular patterns. A team from the University of Barcelona and the Universitat Oberta de Catalunya has now developed a method to make it easier to recognize QR codes in these complex physical settings.
The new approach is not entirely dependent on the surface’s shape and can be used on QR codes found on curved surfaces like bottles and food trays. This is the first technological method that merges a general strategy with two-dimensional barcodes to simplify the recognition of digital data.
This research, published in Pattern Recognition Letters, is led by Professor Ismael Benito from the UB’s Faculty of Physics and the Department of Electronic and Biomedical Engineering, along with the Department of Computer Science, Multimedia, and Telecommunications at UOC. Co-authors include Cristian Fàbrega and Joan Daniel Prades from the Faculty of Physics and the UB’s Institute of Nanoscience and Nanotechnology (IN2UB), as well as experts Hanna Lizarzaburu-Aguilar and David Martínez Carpena from the UB’s Faculty of Mathematics and Computer Science. All contributing authors were involved in various aspects of creating ColorSensing, SL, a UB spin-off focused on smart labeling solutions.
What makes some QR codes hard to scan?
QR codes are an advanced version of traditional barcodes, designed to hold data in a two-dimensional format composed of black and white pixels. They provide quick access to important information, save time, and reduce paper usage, significantly changing how we retrieve information digitally.
However, there can be challenges when scanning a barcode accurately. According to Benito from UB’s Department of Electronic and Biomedical Engineering and a former technology director at ColorSensing, this often happens due to several factors: “First, the quality of the image can be lacking. Many people today have decent digital cameras, yet they can’t always capture a QR code properly. Secondly, the QR code’s print quality and the colors used may not always offer enough contrast. Lastly, if the surface on which the code is printed isn’t adequately flat and parallel to the scanning plane, it complicates the reading process.”
“For instance, these issues arise when attempting to scan a Bicing QR code with a mobile app: the surface is cylindrical, making it challenging to capture the QR code from certain distances—too close (5-10 centimeters) shows distortion, too far (1 meter) makes the code too small. The optimal distance for clear capture is between 30-50 centimeters, where surface distortion is diminished and image quality is sufficient,” explains Benito.
An algorithm that utilizes QR code properties
The research, part of Ismael Benito’s doctoral studies at UB, introduces a novel algorithm that leverages the internal patterns of the QR code to reconstruct the surface it’s on.
The surface’s texture is retrieved using mathematical functions known as splines, which provide local adjustments to traverse surface topographies. Benito mentions, “These functions adjust to the surface’s contours and have applications in geology and photo editing to correct or create surface deformations.”
Numerous technological hurdles still exist in enhancing the QR code recognition process. For commercial applications using a user’s scanner, the key challenge is achieving consistent and accurate readings. “We are also focusing on securing codes against manipulation techniques that could use fake URLs to capture sensitive data. In industrial environments, where the scanning is done under controlled conditions, the primary goal is to speed up the capture process,” he explains.
Founded at UB in 2020 by Professor Joan Daniel Prades and María Eugenia Martín, who is now CEO, ColorSensing received the Metropolitan Business Innovation Award 2023 for developing smart labels aimed at minimizing food waste. The company was also honored with the UB’s Senén Vilaró Award for best innovative company in 2022 and has been recognized in the smart and active packaging category at the Sustainability Awards 2022. Additionally, they hold patents in the United States and Europe, co-promoted by Ismael Benito (UB and UOC), Olga Casals (UB), Cristian Fàbrega (UB), Joan Daniel Prades (UB and Technical University of Braunschweig, TUB), and Andreas Waag (TUB).
