摘要
原木在干燥过程中端面会出现开裂,影响原木质量。为了自动、准确、高效地检测原木的端面裂纹信息,本文提出了一种基于深度学习的原木端面裂纹定量检测方法。分别使用FCN、U-Net、U-Net++三种语义分割网络对原木端面中的裂纹区域及端面区域进行分割。通过识别不同的开裂类型,将分割出的整体裂纹划分成单独裂纹,并利用Hough变换圆检测算法检测出轮裂。通过分割出的裂纹图像和端面图像定义开裂占比、开裂区域面积、端面区域面积,通过裂纹的骨架线长度和裂纹轮廓内的最大内接圆直径描述裂纹的长度和最大宽度。最后分别对裂纹划分前和裂纹划分后进行计算并输出可视化图片。结果表明:U-Net++模型对裂纹的检测效果更优,研究结果可为原木质量评估提供数据支持。
Cracks will appear on the end face of logs during drying process,and the log quality will be affected.In order to automatically,accurately,and efficiently detect the end face crack information of logs,in this paper,a log end face crack quantitative detection method based on deep learning was proposed.FCN,U-Net,and U-Net++semantic segmentation networks were used to segment the crack area and the end area of the log end face.The segmented overall cracks were divided into individual cracks by identifying different crack types,and the Hough transform circle was used to detect ring crack.The crack proportion,crack area,and end face area were defined by the segmented crack image and end surface image.The length and maximum width of the crack were described by the length of the crack skeleton line and the maximum inscribed circle diameter within the crack contour.Finally,visualization images were calculated and output before and after crack division.The test results showed that the U-Net++model has achieved better detection results,the results could provide supportive data for the evaluation of log quality.
作者
李园
郑圣龙
何雨晨
解林坤
周晓剑
杜官本
周华
万辉
LI Yuan;ZHENG Sheng-ong;HE Yu-chen;XIE Lin-kun;ZHOU Xiao-jian;DU Guan-ben;ZHOU Hua;WAN Hui(College of Materials and Chemical Engineering,Southwest Forestry University,Kunming 650233,Yunnan,P.R.China;College of Big Data and Intelligent Engineering,Southwest Forestry University,Kunming 650233,Yunnan,P.R.China)
出处
《林产工业》
北大核心
2024年第4期42-48,58,共8页
China Forest Products Industry
基金
云南省人才培养专项(80201402)。
