Technique for Identification of Electrical Substation Equipment Through Auto-Framing Interest Points and OCR Recognition of Text Tags in Environment for Augmented Reality Systems

• Angel Rodrigues Ferreira Universidade Federal de Uberlândia https://orcid.org/0000-0003-2461-2055
• Jair de Oliveira Pereira Neto Universidade Federal de Uberlândia http://orcid.org/0000-0002-8857-0395
• Alexandre Cardoso Universidade Federal de Uberlândia https://orcid.org/0000-0002-2023-9647
• Edgard Lamounier Junior Universidade Federal de Uberlândia https://orcid.org/0000-0001-6293-9521
• Maurício José Aureliano Júnior Universidade Federal de Uberlândia http://orcid.org/0000-0003-0762-6308
• Diogo Cavalcante de Lima Universidade Federal de Uberlândia https://orcid.org/0000-0001-8956-1768
• Alexandre Carvalho Silva Instituto Federal Goiano https://orcid.org/0000-0003-0264-3475
• Davidson Pereira Campos Eletronorte – Eletrobras https://orcid.org/0009-0004-6472-3256
• João Batista Soares Feitosa Eletronorte – Eletrobras https://orcid.org/0009-0006-9650-4503

Resumo

Electrical power substation systems are considered critical environments with a high impact factor on society. Although Augmented Reality (AR) solutions are becoming increasingly prevalent in the future of Industry 4.0, there is a concern about the practicality of using these systems. AR has significant potential to support assisted maintenance of substation components by projecting field asset information onto an AR headset. To enhance this process, a technique was proposed to identify equipment by automatically reading text from its tags using OCR (generally manual) and using auto-framing through object detection (with Neural Networks). The developed solution can be tested and evaluated in a laboratory setting. The conditions evaluated when pointing a camera at the image of the equipment's operation box with his text identification tag showed that the method employed by the technique can achieve relatively better results than manual framing, making the equipment identification process efficient and potentially promising for implementation in AR devices.

Palavras-chave: Augmented Reality, Industry 4.0, Neural Networks, Object Recognition, OCR

Referências

Sergio Oliveira Frontin. 2013. Equipamentos de Alta Tensão – Prospecção e Hierarquização de Inovações Tecnológicas (1st ed.).

T. Guan and C. Wang. 2009. Registration Based on Scene Recognition and Natural Features Tracking Techniques for Wide-Area Augmented Reality Systems. IEEE Trans Multimedia 11, 8 (December 2009), 1393–1406. DOI: 10.1109/TMM.2009.2032684.

Diego Gouvêa Macharete Trally. 2011. Segmentação de caracteres tipográficos em imagens complexas. Dissertação. Universidade Federal do Rio de Janeiro, Rio de Janeiro.

Tesseract User Manual | tessdoc. Retrieved March 10, 2024 from [link].

Juan Terven, Diana-Margarita Córdova-Esparza, and Julio-Alejandro Romero-González. 2023. A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NAS. Mach Learn Knowl Extr 5, 4 (November 2023), 1680–1716. DOI: 10.3390/make5040083.

Christine Dewi, Rung-Ching Chen, and Hui Yu. 2020. Weight analysis for various prohibitory sign detection and recognition using deep learning. Multimed Tools Appl 79, 43–44 (November 2020), 32897–32915. DOI: 10.1007/s11042-020-09509-x.

Hendry and Rung Ching Chen. 2019. Automatic License Plate Recognition via sliding-window darknet-YOLO deep learning. Image Vis Comput 87, (July 2019), 47–56. DOI: 10.1016/j.imavis.2019.04.007.

Zhiqin Chen, Yufeng Zhang, Hesheng Wang, and Weidong Chen. 2016. Real-time tag recognition based on morphology and local contrast. In 2016 IEEE International Conference on Real-time Computing and Robotics (RCAR), June 2016. IEEE, 614–619. DOI: 10.1109/RCAR.2016.7784100.