UDC 004.932
UNIFIED NEURAL NETWORK METHOD FOR 2D AND 1D BARCODE DETECTION WITH GEOMETRIC DISTORTION CORRECTION
K. P. Lapin, student, Faculty of Computer Engineering, RSREU, Ryazan, Russia;
orcid.org/0009-0004-9279-0739, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
I. I. Sokol, student, Faculty of Computer Engineering, RSREU, Ryazan, Russia;
orcid.org/0009-0001-3539-1596, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
Y. R. Muratov, PhD (in technical sciences), associate professor, Department of Electronic Computers,
RSREU, Ryazan, Russia;
orcid.org/0000-0002-1664-3954, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
A. I. Efimov, PhD (in technical sciences), associate professor, Department of Electronic Computers,
RSREU, Ryazan, Russia;
orcid.org/0000-0002-4014-8718, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The problem of detecting 2D and 1D barcodes in images is considered, allowing for the presence of
multiple codes of different types within a single scene. The aim of this work is to develop a unified neural
network method that provides precise localization of DataMatrix, Aztec, 1D barcodes, QR-Code, PDF417,
and MaxiCode while accounting for variations in imaging conditions, including geometric distortions, nonuniform illumination, and noise on the code surfaces. A lightweight version of the YOLOv8 model is used as
the neural network localizer, enabling deployment in real-time embedded systems with support for neural
accelerators. For 1D barcodes, geometric distortion correction is performed based on gradient analysis. For
DataMatrix and Aztec codes, a similar correction is implemented using an optimized LSD method. Both approaches are aimed at increasing the probability of successful subsequent decoding.
Key words: barcode localization, 2D codes, 1D codes, DataMatrix, QR-Code, Aztec, Barcode,
MaxiCode, PDF417, YOLOv8, LSD, geometric distortion correction