illustration of the three stages in the project; see explanatory text below

Project: Detection and modelling of knots and fibre orientation in sawn timber based on scanning, modelling and machine learning

This research project aims to detect internal properties of sawn timber by integrating CT X-ray and laser scanning. Insights into knot and growth layer geometry, as well as 3D fiber orientation within logs, will serve as the basis for mathematical models. These models provide training data for machine learning networks, enabling us to understand internal wood properties from surface scans alone.

Illustration: Modelling of wood – knots in a log and board and fiber orientation around a knot.
Machine learning – images from models and scanning can be transformed using deep learning networks.
Optical scanner – useful both for research and industry production.

Project information

Project manager
Anders Olsson
Other project members
Osama Abdeljaber, Min Hu
Participating organizations
Linnaeus University, Microtec, Ikea, Södra
Financier
The KK Foundation
Timetable
1 July 2022–30 June 2026
Subject
Building technology (Department of Building Technology, Faculty of Technology)

More about the project

The research project will contribute to the development of more accurate detection of characteristics in sawn timber and wood products than can be done today. Not through step-by-step improvements of already established methods but by using a completely new method, where optical scanning and modelling of wood material is combined with applied machine learning.

Collaboration between the project group at Linnaeus University and the company Microtec – and interconnection of expertise in wood science, machine learning and mathematical modelling of wood – have already led to a novel method to determine pith location along timber boards, and to two patent applications based on this idea. In this new project, the same innovative idea will be developed and used for other and more challenging applications, such as more precise detection of knots and fibre orientation inside timber boards of different species. Participating companies in addition to Microtec are Ikea and Södra. The latter represent typical producers who will benefit from more accurate detection and grading of wood material.

For the research environment Wood Building Technology at Linnaeus University and especially the project group, who have extensive experience of modelling and assessment of wood, the proposed project represents a significant development. This is because machine learning through the project is expected to become an integrated tool in its applied research on wood and wood-based products.

Staff