浏览全部资源
扫码关注微信
南京林业大学材料科学与工程学院，江苏南京 210037
纸质出版日期： 2024-01-30 ，
扫 描 看 全 文
洪吾俊,李万兆,胡尧琼等.基于卷积神经网络的刨花定向角度自动测量方法构建[J].木材科学与技术,2024,38(01):58-65.
HONG Wujun,LI Wanzhao,HU Yaoqiong,et al.An Automatic Method for Measuring Strands’ Orientation Angles Based on Convolutional Neural Network[J].Chinese Journal of Wood Science and Technology,2024,38(01):58-65.
基于卷积神经网络YOLOv5和最小外接矩形算法，构建一种自动准确地采集铺装刨花定向角度的方法。结果表明，构建的YOLOv5模型识别刨花目标的准确率、召回率和
F
1值分别为0.992、0.897和0.94，能够有效识别层叠刨花。模型自动测量和人工测量的刨花定向角度具有强相关性（
R
2
=0.99），且模型不存在算法缺陷，计算每张刨花铺装图像（像素640×640）用时仅134.7 ms。该刨花定向角度计算模型可以为工业领域优化OSB生产工艺以及提高产品性能提供技术支撑。
Oriented strand board (OSB) is a popular wood-based panel product that has gained widespread application in construction
furniture
and other industries due to the high strength
durability
and cost-effectiveness. The basic component of OSB is large strand that is arranged in a specific orientation and bonded together using adhesives to form a strong and durable panel product. The strand’s orientation angle is a key factor that affects the mechanical properties of OSB. The effective collection of information on strands’ orientation angles is important for optimizing the production process and improving product performance. In this study
a method was developed for automatically measuring the strand’s orientation angle with computer vision and machine learning techniques. The method is based on the convolutional neural network YOLOv5 and the minimum outer rectangle algorithm
which can accurately and efficiently identify and measure the strand’s orientation angle during the stranding process. The study evaluated the performance of YOLOv5
the accuracy of automatically calculating the strand’s orientation angle
and the time efficiency of the automatic strand’s orientation angle acquisition. The results of the study showed that the YOLOv5 model was effective in identifying the object from the images of the layered strands
with an accuracy
recall
and
F
1 value of 0.992
0.897
and 0.94. The model can effectively identify the layered strands. The orientation angles of strands measured by automatic and manual methods were strongly correlated (
R
2
= 0.99). There were no orientation angle algorithmic flaws. Furthermore
the method established in this study has high time efficiency
with a processing time of only 134.7 ms for each 640×640-pixel image. In conclusion
the method for automatically measuring the strand’s orientation angle using YOLOv5 and minimum outer rectangle algorithm provides an efficient and accurate way to collect strands’ orientation angle information during the stranding process
which is crucial for optimizing the production process and improving product performance. This research has important practical applications in the wood-based panel industry and provides a valuable reference for future research in this area.
定向刨花板卷积神经网络刨花识别定向角度计算模型性能评价
oriented strand boardconvolutional neural networkstrand identificationorientation angle calculationmodel performance evaluation
李万兆, 李东虎, 陈超意, 等. 定向刨花板和细表面定向刨花板抗弯性能及应变分布比较[J]. 林业工程学报, 2022, 7(6): 61-66.
LI W Z, LI D H, CHEN C Y, et al. Comparison of bending strength and strain distribution in oriented strand board and fine surface oriented strand board[J]. Journal of Forestry Engineering, 2022, 7(6): 61-66.
胡尧琼, 李万兆, 张晔, 等. 刨花形态和跌落高度对刨花定向效果的影响[J]. 林业工程学报, 2023, 37(4): 66-71.
HU Y Q, LI W Z, ZHANG Y, et al. Effect of strand shape and drop height on strands orientation[J]. Journal of Forestry Engineering, 2023, 37(4): 66-71.
NISHIMURA T, ANSELL M P, ANDO N. The relationship between the arrangement of wood strands at the surface of OSB and the modulus of rupture determined by image analysis[J]. Wood Science and Technology, 2001, 35(6): 555-562.
彭明凯, 梅长彤. 刨花形态及定向角度对刨片层积材性能的影响[J]. 南京林业大学学报(自然科学版), 2009, 33(4): 129-131.
PENG M K, MEI C T. Effects of strand geometry and orientation on the properties of laminated strand lumber[J]. Journal of Nanjing Forestry University (Natural Sciences Edition), 2009, 33(4): 129-131.
NISHIMURA T, AMIN J, ANSELL M P. Image analysis and bending properties of model OSB panels as a function of strand distribution, shape and size[J]. Wood Science and Technology, 2004, 38(4): 297-309.
梅长彤, 周又平, 薛联凤. 刨花形态自动识别系统的设计与开发[J]. 林业科技开发, 2006, 20(3): 49-51.
MEI C T, ZHOU Y P, XUE L F. A design for flake from auto recognition system and its exploitation[J]. Journal of Forestry Engineering, 2006, 20(3): 49-51.
吕同玲, 任洪娥, 马岩. 长条薄片刨花视频检测方法的探讨[J]. 木材加工机械, 2007, 18(4): 19-21.
LYU T L, REN H E, MA Y. Study on the slice video frequency examination method[J]. Forestry and Grassland Machinery, 2007, 18(4): 19-21.
杜春贵, 陈思果, ROBERT Wellwood. 定向刨花板定向角的激光测量[J]. 光电子．激光, 2011, 22(5): 744-747.
DU C G, CHEN S G, ROBERT W. Measurement of alignment angle for oriented strand board using laser technology[J]. Journal of Optoelectronics．Laser, 2011, 22(5): 744-747.
NYSTRÖM J. Automatic measurement of fiber orientation in softwoods by using the tracheid effect[J]. Computers and Electronics in Agriculture, 2003, 41(1/2/3): 91-99.
DONG C, LOY C C, HE K M, et al. Image super-resolution using deep convolutional networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2): 295-307.
YANG S Y, KWON O, PARK Y, et al. Application of neural networks for classifying softwood species using near infrared spectroscopy[J]. Journal of Near Infrared Spectroscopy, 2020, 28(5/6): 298-307.
肖雨晴, 杨慧敏, 王柯欣, 等. 卷积神经网络在木材缺陷检测应用中的研究进展[J]. 木材科学与技术, 2021, 35(3): 12-18.
XIAO Y Q, YANG H M, WANG K X, et al. Research progress of convolutional neural network in wood defect detection[J]. Chinese Journal of Wood Science and Technology, 2021, 35(3): 12-18.
杨攀, 郑积仕, 冯芝清, 等. 基于Mask R-CNN的密集木材检测分割方法[J]. 林业工程学报, 2022, 7(2): 135-142.
YANG P, ZHENG J S, FENG Z Q, et al. Research on detection and segmentation methods of dense-stacked logs using Mask R-CNN[J]. Journal of Forestry Engineering, 2022, 7(2): 135-142.
谢文博. 基于改进Mask R-CNN的木材缺陷检测分割算法研究[D]. 哈尔滨: 哈尔滨理工大学, 2022.
席靖宇, 王宇轩, 衡利辰, 等. 基于计算机视觉的针叶材木射线特征提取方法[J]. 林业工程学报, 2023, 8(3): 132-140.
XI J Y, WANG Y X, HENG L C, et al. Wood ray feature extraction method of softwood based on computer vision[J]. Journal of Forestry Engineering, 2023, 8(3): 132-140.
REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.
HE K M, GKIOXARI G, DOLLAR P, et al. Mask R-CNN[C]//2017 IEEE International Conference on Computer Vision (ICCV), 2017: 2961-2969.
HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 770-778.
HE K M, ZHANG X Y, REN S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9): 1904-1916.
EBERLY D. Minimum-area rectangle containing a set of points[J/OL]. Geometric Tools, LLC, 2015.https://www.geometrictools.com/Documentation/MinimumAreaRectangle.pdfhttps://www.geometrictools.com/Documentation/MinimumAreaRectangle.pdf.
付美子, 林炳清. 临床测量中定量数据Bland-Altman一致性评价[J]. 广西师范大学学报(自然科学版), 2022, 40(1): 125-138.
FU M Z, LIN B Q. Bland-Altman method for assessing agreement of quantitative data in clinical measurement[J]. Journal of Guangxi Normal University (Natural Science Edition), 2022, 40(1): 125-138.
BOLYA D, ZHOU C, XIAO F Y, et al. YOLACT: real-time instance segmentation[C]//2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019: 9157-9166.
REDMON J, FARHADI A. Yolov3: An incremental improvement[J/OL]. ArXiv preprint arXiv:1804.02767, 2018. https://arxiv.org/abs/1804.02767https://arxiv.org/abs/1804.02767.
RIFFENBURGH R H. Statistics in medicine[M]. 3rd ed. Oxford: Academic Press, 2012: 631-649.
相关作者
相关机构
京公网安备11010802024621
微信公众号