基于树木年轮径切特征的卷积神经网络树种识别

doi:10.13287/j.1001-9332.202301.001

应用生态学报 ›› 2023, Vol. 34 ›› Issue (1): 47-57.doi: 10.13287/j.1001-9332.202301.001

基于树木年轮径切特征的卷积神经网络树种识别

高欣¹, 杨立新^1*, 陈振举^1,2,3,4

¹沈阳农业大学林学院, 树木年轮实验室/辽宁辽河平原森林生态系统国家定位观测研究站, 沈阳 110866;
²中国科学院清原森林生态系统观测研究站, 沈阳 110164;
³中国科学院沙漠与沙漠化重点实验室, 兰州 730000;
⁴吉林长白山森林生态系统国家野外观测研究站, 吉林二道白河 133613

收稿日期:2022-08-11 修回日期:2022-10-11 出版日期:2023-01-15 发布日期:2023-06-15
通讯作者: ^*E-mail: 285277489@qq.com
作者简介:高欣, 女, 1998年生, 硕士研究生。主要从事景观生态、人工智能和树木年轮学研究。E-mail: 760321531@qq.com
基金资助:
国家自然科学基金项目( 41871027,41888101,31570632) 资助。

Convolutional neural network tree species identification based on tree-ring radial section image features

GAO Xin¹, YANG Li-xin^1*, CHEN Zhen-ju^1,2,3,4

¹Tree-Ring Laboratory/Research Station of Liaohe-River Plain Forest Ecosystem CFERN, College of Forestry, Shenyang Agricultural University, Shenyang 110866, China;
²Qingyuan Forest CERN, Chinese Academy of Sciences, Shenyang 110164, China;
³Key Laboratory of Desert and Desertification, Chinese Academy of Sciences, Lanzhou 730000, China;
⁴National Research Station of Changbai Forest Ecosystem, Er'daobaihe 133613, Jilin, China

Received:2022-08-11 Revised:2022-10-11 Online:2023-01-15 Published:2023-06-15

摘要/Abstract

摘要： 卷积神经网络可以通过树木年轮样本构造特征图像实现物种识别的自动化。本研究通过建立树木年轮样本构造特征图像集,选用LeNet、AlexNet、GoogLeNet和VGGNet 4个卷积神经网络模型,实现基于树木年轮横切面的计算机自动化树种精准识别,进而确定各模型的树种识别准确率,明晰不同树种在自动识别中的混淆情况,探测不同模型识别结果的差异。结果表明: 本研究训练的用于树种识别的卷积神经网络模型具有较好的可信度;4个模型中GoogLeNet模型树种识别准确率最高,为96.7%,LeNet模型识别准确率最低(66.4%);不同模型对于所选树种的识别结果具有一致性,表现为对蒙古栎识别准确率最高(AlexNet模型识别率达到100%),对臭冷杉的识别准确率最低。本研究中也存在类似结构树种的识别混淆情况。模型在科和属水平的识别准确率高于种水平;阔叶树种因其显著的结构差异容易区分,阔叶树树种的识别准确率高于针叶树。总体上,通过卷积神经网络,探测了树木年轮特征的深层信息,达到树种的精准识别,提供了一种快速便捷的自动树种初筛鉴定方法。

关键词: 树种识别, 树木年轮, 径切, 卷积神经网络

Abstract: Convolutional neural networks can automatically identify tree species based on the images of structural features of tree-rings samples. In this study, we used a tree-ring image dataset for different species to achieve tree-ring based species automatic identification with high accuracy by four convolutional neural network models (LeNet, AlexNet, GoogLeNet, and VGGNet), aiming to determine the identification accuracy of the models, clarify the species misidentification during the automatic processes, and explore the identification differences among the models. The results showed that tree species identification derived from the trained convolutional neural network models was reliable, with the GoogLeNet and LeNet showed the highest (96.7%) and lowest (66.4%) identification accuracy. The tree species identifications using different models were highly consistent. Quercus mongolica and Abies nephrolepis showed the highest (100% for AlexNet) and lowest identification accuracy, respectively. Misidentification could occur among tree species with similar tree-ring structure. The identification accuracy of the models was higher at family and genus levels than that at the species level. The identification accuracy of broadleaf tree species was higher than that of coniferous trees due to distinct radial structure among broadleaf species. Overall, our method achieved a high accuracy for tree species identification, and provided a fast, convenient, and automatic tree species identification by detecting specific tree-ring properties with convolutional neural network.

Key words: tree species identification, tree ring, radial section, convolutional neural network.

高欣, 杨立新, 陈振举. 基于树木年轮径切特征的卷积神经网络树种识别[J]. 应用生态学报, 2023, 34(1): 47-57.

GAO Xin, YANG Li-xin, CHEN Zhen-ju. Convolutional neural network tree species identification based on tree-ring radial section image features[J]. Chinese Journal of Applied Ecology, 2023, 34(1): 47-57.

参考文献

[1] 何拓, 焦立超, 郭娟, 等. 木材信息学: 发展、应用与展望. 木材科学与技术, 2021, 35(4): 15-24
[2] 方克艳, 陈秋艳, 刘昶智, 等. 树木年代学的研究进展. 应用生态学报, 2014, 25(7): 1879-1888
[3] 李俊霞, 马鑫博, 付永平, 等. 沈阳清墓木质葬具的树种鉴定及生态环境的指示潜力. 应用生态学报, 2021, 32(10): 3687-3698
[4] Jin YT, Li JX, Bai XP, et al. High temperatures constrain latewood formation in Larix gmelinii xylem in boreal forests. Global Ecology and Conservation, 2021, 30: e01767
[5] 赵莹, 蔡立新, 靳雨婷, 等. 暖干化加剧东北半干旱地区油松人工林径向生长的水分限制. 应用生态学报, 2021, 32(10): 3459-3467
[6] Homan D, Preez JA. Automated feature-specific tree species identification from natural images using deep semi-supervised learning. Ecological Informatics, 2021, 66: 101475
[7] Fabijańska A, Danek M, Barniak J. Wood species automatic identification from wood core images with a resi-dual convolutional neural network. Computers and Electronics in Agriculture, 2021, 181: 105941
[8] Ravindran P, Costa A, Soares R, et al. Classification of CITES-listed and other neotropical Meliaceae wood images using convolutional neural networks. Plant Methods, 2018, 14: 25
[9] Lens F, Liang C, Guo YH, et al. Computer-assisted timber identification based on features extracted from microscopic wood sections. IAWA Journal, 2020, 41: 660-680
[10] 张怡卓, 马琳, 王铁滨, 等. 小波变换的木材纹理在线分选. 林业科技, 2012, 37(6): 21-24
[11] Filho P, Oliveira LS, Nisgoski S, et al. Forest species recognition using macroscopic images. Machine Vision and Applications, 2014, 25: 1019-1031
[12] Kobayashi K, Hwang SW, Lee WH, et al. Texture ana-lysis of stereograms of diffuse-porous hardwood: Identification of wood species used in Tripitaka Koreana. Journal of Wood Science, 2017, 63: 322-330
[13] Peng Z. Robust wood species recognition using variable color information. Optik, 2013, 124: 2833-2836
[14] Zamri MIP, Cordova F, Khairuddin ASM, et al. Tree species classification based on image analysis using Improved-Basic Gray Level Aura Matrix. Computers and Electonics in Agriculture, 2016, 124: 227-233
[15] Yusof R, Rosli NR. Tropical wood species recognition system based on Gabor filter as image multiplier. 9th International Conference on Signal-Image Technology and Internet-Based Systems, Kyoto, 2013: 737-743
[16] Barmpoutis P, Dimitropoulos K, Barboutis I, et al. Wood species recognition through multidimensional texture analysis. Computers and Electronics in Agriculture, 2018, 144: 241-248
[17] Geus A, Backes AR, Gontijo AB, et al. Amazon wood species classification: A comparison between deep lear-ning and pre-designed features. Wood Science and Technology, 2021, 55: 857-872
[18] Wu FY, Gazo R, Haviarova E, et al. Wood identification based on longitudinal section images by using deep learning. Wood Science and Technology, 2021, 55: 553-563
[19] Lecun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition. Proceedings of the IEEE, 1998, 86: 2278-2324
[20] Szegedy C, Liu W, Jia YQ, et al. Going deeper with convolutions. IEEE Conference on Computer Vision and Pattern Recognition, Boston, MA, USA, 2015: 1-9
[21] Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. Communications of the ACM, 2017, 60: 84-90
[22] Karpathy A, Toderici G, Shetty, et al. Large-scale video classification with convolutional neural networks. IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA, 2014: 1725-1732
[23] Hinton GE, Srivastava N, Krizhevsky A, et al. Improving neural networks by preventing co-adaptation of feature detectors. Computer Science, 2012, 3: 212-223
[24] Nair V, Hinton GE. Rectified linear units improve restricted Boltzmann machines Vinod Nair. Proceedings of the 27th International Conference on Machine Learning, Haifa, Israel, 2010: 807-814
[25] Passalis N, Tefas A. Learning bag-of-features pooling for deep convolutional neural networks. IEEE International Conference on Computer Vision, Venice, Italy, 2017: 5766-5774
[26] Kingma DP, Ba J. A method for stochastic optimization. Proceddings of the 3rd International Conference for Learning Representations, San Diego, CA, USA, 2015: 1-15
[27] Smith LN, Topin N. Super-convergence: Very fast trai-ning of neural networks using large learning rates. Conference on Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications, Baltimore, MD, USA, 2019: UNSP1100612
[28] 杨家驹, 程放, 杨建华, 等. 木材识别: 主要乔木树种. 北京: 中国建材工业出版社, 2009: 188-193
[29] 刘一星, 赵广杰. 木质资源材料学. 北京: 中国林业出版社, 2004: 65-72
[30] 徐有明. 木材学. 第2版. 北京: 中国林业出版社, 2019: 76-82
[31] Elmas B. Identifying species of trees through bark images by convolutional neural networks with transfer lear-ning method. Journal of the Faculty of Engineering and Architecture of Gazi University, 2021, 36: 1254-1269
[32] Vizcarra G, Bermejo D, Mauricio A, et al. The Peru-vian Amazon forestry dataset: A leaf image classification corpus. Ecological Informatics, 2021, 62: 101268
[33] Minowa Y, Kubota Y. Identification of broad-leaf trees using deep learning based on field photographs of multiple leaves. Journal of Forest Research, 2022, 27: 246-254
[34] Szegedy C, Vanhoucke V, Ioffe S, et al. Rethinking the inception architecture for computer vision. IEEE Confe-rence on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA, 2016: 2818-2826
[35] 何拓, 刘守佳, 陆杨, 等. iWood: 基于卷积神经网络的濒危珍贵树种木材自动识别系统. 林业科学, 2021, 57(9): 152-159
[36] Figueroa-Mata G, Mata-Montero E, Valverde-Otarola JC, et al. Using deep learning to identify Costa Rican native tree species from wood cut images. Frontiers in Plant Science, 2022, 13: 789227
[37] 殷笑寒, 郝广友. 长白山阔叶树种木质部环孔和散孔结构特征的分化导致其水力学性状的显著差异. 应用生态学报, 2018, 29(2): 352-360
[38] Gasson P. How precise can wood identification be? Wood anatomy's role in support of the legal timber trade, especially cites. IAWA Journal, 2011, 32: 137-154
[39] da Silva NR, de Ridder M, Baetens JM, et al. Automated classification of wood transverse cross-section micro-imagery from 77 commercial Central-African timber species. Annals of Forest Science, 2017, 74: 30
[40] Wheeler EA, Baas P. Wood identification: A review. IAWA Journal, 1998, 19: 241-264

基于树木年轮径切特征的卷积神经网络树种识别

Convolutional neural network tree species identification based on tree-ring radial section image features

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	孙龙, 马灵感, 郭妍, 范佳乐, 陈伯轩, 胡同欣. 基于卷积神经网络及气象要素回归法的地表死可燃物含水率预测模型 [J]. 应用生态学报, 2023, 34(9): 2453-2461.
[2]	李颖辉, 齐贵增, 冯荣荣, 刘康. 秦岭北麓油松径向生长对气候变化的响应 [J]. 应用生态学报, 2022, 33(8): 2043-2050.
[3]	韩艳刚, 盖学瑞, 邱思玉, 张月, 王守乐, 周莉, 于大炮. 大兴安岭兴安落叶松径向生长对气候响应的时空变化 [J]. 应用生态学报, 2021, 32(10): 3397-3404.
[4]	杨婧雯, 张秋良, 宋文琦, 张旭, 李宗善, 张远东, 王晓春. 大兴安岭兴安落叶松和樟子松径向生长对气候变化的响应差异 [J]. 应用生态学报, 2021, 32(10): 3415-3427.
[5]	肖健宇, 张文艳, 牟玉梅, 吕利新. 树木年轮揭示的东灵山主要树种间干旱耐受性差异 [J]. 应用生态学报, 2021, 32(10): 3487-3496.
[6]	李静茹, 彭剑峰, 杨柳, 崔佳月, 李轩, 彭猛. 川西高原两种针叶树径向生长对气候因子的响应 [J]. 应用生态学报, 2021, 32(10): 3512-3520.
[7]	刘兰娅, 勾晓华, 张芬, 尹定财, 王学佳, 夏敬清, 李芊, 杜苗苗. 升温对祁连山东部青海云杉径向生长的影响 [J]. 应用生态学报, 2021, 32(10): 3576-3584.
[8]	苟晓霞, 张同文, 袁玉江, 喻树龙, 张瑞波, 姜盛夏, 郭玉琳. 阿尔泰山主要针叶树种树木径向生长及其对气候变化的响应 [J]. 应用生态学报, 2021, 32(10): 3594-3608.
[9]	曹红华, 赵晓恩, 陈峰, 王世杰, 刘兴华. 基于油松树轮重建晋东南地区近296年来1—6月降水变化 [J]. 应用生态学报, 2021, 32(10): 3618-3626.
[10]	袁晴雪, 喻树龙, 姜盛夏, 秦宁生, 张同文. 川西雅江地区过去567年3—8月干旱变化的树轮重建 [J]. 应用生态学报, 2021, 32(10): 3627-3635.
[11]	朱晓龙, 陈庆美, 吕爱锋, 黄茹, 邵雪梅, 梁尔源, 朱海峰. 利用树轮揭示人类活动对柴达木盆地巴音河径流量的影响 [J]. 应用生态学报, 2021, 32(10): 3653-3660.
[12]	白耀玉, 韩宜洁, 王坷坷, 刘波. 秦岭山地针叶树种树木生长对气候变化的响应 [J]. 应用生态学报, 2021, 32(10): 3715-3723.
[13]	康虎虎, 刘晓宏, 张馨予, 郭军明, 吴国菊, 徐国保, 康世昌. 树木年轮汞记录: 进展、问题和展望 [J]. 应用生态学报, 2021, 32(10): 3733-3742.
[14]	李瑶, 曾小敏, 倪萍, 王文超, 刘晓宏. 树木年轮年内高分辨率稳定同位素记录:方法、进展和展望 [J]. 应用生态学报, 2021, 32(10): 3743-3752.
[15]	于健, 陈佳佳, 孟盛旺, 周华, 周光, 高露双, 王永平, 刘琪璟. 长白山群落交错带长白松和鱼鳞云杉径向生长对气候变暖的响应 [J]. 应用生态学报, 2021, 32(1): 46-56.