木材科学与技术2025，Vol.39Issue(3)：49-57,9.DOI:10.12326/j.2096-9694.2025026
基于计算机视觉的定向刨花板内部三维结构信息挖掘方法构建与应用
Construction and Application of Three-Dimensional Structural Data Mining Method for OSB Based on Computer Vision
摘要
Abstract
Oriented strand board(OSB)is a wood-based structural panel,which is primarily composed of long strands.The physical and mechanical properties are closely related to its internal structure.Since the internal structure of OSB is formed through directional alignment and random layered overlapping of strands,precise characterization of the internal structure is crucial for understanding its structure-property relationships.This study proposes a computer vision-based method for constructing the three-dimensional internal structure of OSB.Strands were prepared by cutting peeled veneer sheets,and 1.2 cm thick OSB panels were fabricated.The mat-forming process was continuously recorded using imaging technology.Computer vision-based techniques were employed to capture strand positions after forming,followed by individual strand contour extraction.Combined with vertical density profile(VDP)data for calculating thickness-direction compression ratios,Python software was utilized to reconstruct the 3D internal structure of OSB.Specimens(5 cm×5 cm)were prepared to measure weight,VDP,3D structure,and porosity.Results showed that the total number of strands in fabricated OSB was 2 299,with all strands correctly identified and 2 232 strands(97.1%)completely segmented through computer processing.Determination coefficient(R2)between measured and predicted values reached 0.976 for weight,0.951 for porosity,and 0.812 for VDP respectively.Comparative analysis with X-ray computed tomography(CT)showed excellent consistency in 3D visualization.Vision-based construction enables quantitative analysis of orientation angles,surface areas,and resin distribution for all strands.This computer vision approach provides exact visualization and quantification of OSB's internal 3D structure,establishing reliable fundamental data for investigating structure-property relationships in OSB.关键词
定向刨花板/计算机视觉/断面密度梯度/三维结构/准确性评价Key words
oriented strand board/computer vision/vertical density profile/three-dimensional structure/accuracy evaluation分类
轻工纺织引用本文复制引用
石允峰,洪吾俊,段衍筠,李万兆,梅长彤..基于计算机视觉的定向刨花板内部三维结构信息挖掘方法构建与应用[J].木材科学与技术,2025,39(3):49-57,9.基金项目
"十四五"国家重点研发计划项目课题"高强定向结构板制造与应用技术"(2021YFD2200602). （2021YFD2200602）
