Dyeing adhesives as binders in wood/bamboo biocomposites for improve the measurement accuracy of wood failure percentage by machine vision technology

Abstract

Wood failure percentage (WFP) is an important evaluation index for assessing the bonding performance of biocomposites, but it is challenging to measure WFP accurately using machine vision technologies (MVTs). The purpose of this paper is to enhance measurement accuracy by modifying the hue properties of adhesives. Additionally, the physical and chemical changes induced in the adhesives by these dyes were investigated. The results indicated that the strategy significantly improved the accuracy of the three MVTs in measuring WFP, with the order of deep learning (DL) > machine learning (ML) > pixel statistics (PS) and a maximum of relative error decrease of 96.6 %. Furthermore, the strategy notably increased the pixel value of adhesives in the glulam failure surface image, reaching up to 170. Importantly, the strategy had no significant effect on bonding strength and failure mode, as the adhesive range was only 0.13 when compared to the shear state and specimen type. However, the strategy increased the content of Ca and Mg in the adhesives to 12.22 % and 6.32 %, respectively. Overall, this strategy provides a technical means to enhance gluing quality detection for glulam.