Dynamic and intelligent optimization of the data matrix part reading process

• Autor(a) principal: Soares, Paulo Miguel Meneses
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/41002
• topic: Traceability; Datamatrix; Computer vision; Machine learning; Object detection

Ensuring accurate traceability in manufacturing processes is crucial for quality control and product safety. The use of datamatrix codes has become common for encoding essential information. However, challenges arise when decoding these codes due to factors such as poor marking, image noise, and varying lighting conditions. This thesis addresses the problem of reliable datamatrix reading using computer vision and machine learning techniques. The aim of this thesis is to achieve a misreading percentage below 2% by implementing an intelligent system that can accurately decode datamatrix codes. Through extensive research and experimentation, a dynamic solution was developed that leverages image analysis, processing, and optimization algorithms. By applying techniques such as rotation, cropping, and binarization, the system enhances the readability of datamatrix codes and removes extraneous noise. The proposed solution was designed and implemented specifically for quality control in the manufacturing processes of differential boxes at Renault Cacia. The system’s performance was validated through rigorous testing, and the desired goal was surpassed with a reading decoding accuracy of 100%. The system’s implementation was facilitated by the use of computer vision and machine learning principles. The successful integration of intelligent algorithms highlights the potential for further advancements in quality monitoring and real-time analysis within the manufacturing industry.