Visual Stress Grading Automation Using Image Processing and Segmentation Analysis

Author's Department

Construction Engineering Department

Second Author's Department

Construction Engineering Department

Third Author's Department

Construction Engineering Department

Fourth Author's Department

Construction Engineering Department

Fifth Author's Department

Construction Engineering Department

Find in your Library

https://doi.org/10.1007/978-3-031-61531-3_14

All Authors

Bassel Abdel Shahed, Salma Alnaas, Mira Khayrat, Sherif Ihab, Mohamed Darwish, Khaled Nassar, Ezzeldin Sayed-Ahmed

Document Type

Research Article

Publication Title

Lecture Notes in Civil Engineering

Publication Date

1-1-2024

doi

10.1007/978-3-031-61531-3_14

Abstract

The variability in wood mechanical properties is one of the concerning factors when considering timber in structural applications. This variability is influenced by the presence of visible defects such as knots, grain deviations, and splits. Multiple models were developed to predict the mechanical performance of timber by means of visual stress grading. Stress modification factors are determined according to the frequency of knot sizes and slope of the grain within a certain stress grade of a wood species, to be applied on clear wood strength values for that species. The development of stress grades requires large surveys of knot properties and distribution within a timber species, where knot sizes on nominal dimension lumber faces are measured to develop knot data, and these data are used to determine the average sum of knot sizes in 1-foot lengths taken at 2-inch intervals on each timber board. According to the American standards, to develop a stress grade, physical mapping and measurements of knot data for at least 1000 linear foot of lumber should be done. Such exhaustive and time-consuming process can be automated by state-of-the-art computer vision and segmentation analysis techniques. Images are captured for pieces of lumber, and image adjustments are made to enhance contrast and emphasize features. Then, a first-order Gaussian derivative filter is applied on each picture to develop a binary contour image that contains the edge features of all knots. Those components formed by edge detection are then measured in pixels, where the nominal dimension of the lumber is used to set the scale for pixel dimensions to real-life dimensions conversion. This paper purposes a knot detection and segmentation algorithm for Casuarina glauca lumber, resulting in a fully automated knot data collection process.

First Page

177

Last Page

185

Comments

Conference Paper. Record derived from SCOPUS.

Link to Full Text
Find this item online

Share

COinS