Digital image processing in the determination of glued wood joints behavior

The determination of the efficiency of adhesives applied in glued joints, through the analysis of the percentage of wood failure, is one of the main validation parameters for the quality of gluing techniques and adhesive. However, the subjectivity of the interpretation, performance, and visual acuity of the evaluators makes the methodology subjective. This research aimed to establish a method for evaluating the behavior of the glue line in glued joints, using freely available software and digital camera images. Shear samples of glued joints with polyvinyl acetate - PVAc, resorcinol formaldehyde - RF, and melamine formaldehyde - MF were evaluated using digital images obtained by a 16.1-megapixel camera. The images were pre-processed to enable vectorization in Inkscape software, version 1.1.2. Each image was vectorized and subjected to the measurement of the areas of the generated vector layers (glue line and wood). The validation process was performed by evaluating the accuracy of vectorization. Thus, the photograph of each sample and the vectorized image were imported into ImageJ software, version 1.53. To ensure the reliability of the proposed methodology, statistical analyses of the Kappa coefficient, global accuracy, user, and producer were applied. To evaluate the efficiency of the proposed method in relation to the conventional method provided in ASTM D5266-13 , five evaluators analyzed the proportion of bond line and wood failure of the 10 samples submitted to the proposed methodology. Subsequently, a descriptive analysis of the data was carried out to evaluate the variability and subjectivity of the conventional analyses among the evaluators. It is noted that the proposed methodology achieved high efficiency in the analysis of the proportion of glue line and wood, and it can be assured by the accuracy tests performed. The research also showed that there was no agreement among the five evaluators using the conventional method. It is concluded that the proposed methodology meets the research objective, with high levels of accuracy, and confirms the subjectivity of the analyses performed with the conventional methodology. The results underscore the need for objective methodologies, employing computational techniques to reduce subjectivity in adhesive bond failure assessments, enhancing critical decisions in the selection of adhesives for industrial structural applications.