Your new experience awaits. Try the new design now and help us make it even better

ORIGINAL RESEARCH article

Front. Mater.

Sec. Computational Materials Science

Volume 12 - 2025 | doi: 10.3389/fmats.2025.1635222

Detection of wood grain defects based on edge prior aggregation

Provisionally accepted
Weijian  SunWeijian Sun1*Xu  ChengXu Cheng2Sai  WangSai Wang3Tuan-Tuan  WangTuan-Tuan Wang3
  • 1Jiangnan University, Wuxi, China
  • 2Fujian University of Technology, Fuzhou, China
  • 3Hainan University, Haikou, China

The final, formatted version of the article will be published soon.

As a widely distributed renewable resource in nature, wood plays a vital role in accelerating urbanisation. The detection of wood grain defects can avoid the safety hazards they create.However, the original wood texture images have the problems of low clarity and low contrast, which make it difficult to distinguish the boundary information of defects from the original images.At the same time, the defective and non-defective regions of wood texture are similar in colour, making it difficult for the algorithm to distinguish between them, resulting in the phenomenon of missed detection. In this paper, based on the characteristics of wood images and defective textures, a detection network based on edge priori enhancement is proposed to achieve defective texture detection by making full use of edge priori knowledge. Among them, global edge priori enhancement is used to extract the positional relationship between the defective texture and the neighbouring pixels from a large sensory field as a way to capture the key contextual information and make the algorithm more focused on the foreground object. Local edge priori enhancement constructs similarity mappings from multiple feature spaces, highlights local regions with important edge information, and helps the algorithm to identify relevant edge priors from multiple superimposed defects. On publicly available datasets, detection performance reached 0.869 AP50 on wood grain data with single grain defects and 0.914 AP50 on wood data with multiple grain defects. The designed network achieved at least 16.8% improvement in detection performance based on baseline. Sufficient experiments confirm that our algorithm outperforms existing texture defect detection algorithms on publicly available data on wood texture defects, and exhibits superior robustness when dealing with multiple noises.

Keywords: Wood defect, Edge priori knowledge, convolutional neural network (CNN) model, detection performance, Image processing robustness

Received: 26 May 2025; Accepted: 05 Aug 2025.

Copyright: © 2025 Sun, Cheng, Wang and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

* Correspondence: Weijian Sun, Jiangnan University, Wuxi, China

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.