|
基于DenseNet模型的无人机高分辨率影像树种分类研究
|
Abstract:
如今深度学习广泛应用于医学、工业、人工智能以及地理学等领域。本文基于DenseNet模型,在其残差块之间加入1 × 1的小型卷积核作为瓶颈层得到了一种改进的DenseNet_BL模型,以琅琊山林场为研究区,使用DenseNet121_BL和DenseNet169_BL模型对研究区的无人机高分辨率光学影像进行分类研究实验。得到的实验结果表明DenseNet121_BL模型在进行树种分类时正确率最高,达到了88.29%。说明改进后的DenseNet_BL模型是一种有效的树种分类算法。
Deep learning is widely used in medicine, industry, artificial intelligence, geography and other fields. This paper proposes an improved DenseNet_BL model based on DenseNet model. An improved DenseNet_BL model is obtained by adding a 1 × 1 small convolution kernel between the Residual Blocks as the Bottleneck Layer. Taking Langya Mountain Forest as the research area, DenseNet 121_BL and DenseNet169_BL models were used to classify UAV high-resolution optical images in the research area. The experimental results showed that DenseNet_BL121 model had the highest accuracy in tree species classification, reaching 88.29%. The improved DenseNet_BL model is an effective tree species classification algorithm.
[1] | 黎庆华. 林业生态保护存在的问题及对策[J]. 乡村科技, 2020, 11(35): 60-61. |
[2] | 李丹, 黄钰辉, 孙中宇, 张卫强, 甘先华, 王佐霖, 孙红斌, 杨龙. 基于机器学习的深圳市坝光湿地园树种高光谱分类[J]. 红外, 2019, 40(7): 47-52. |
[3] | 吴恒, 朱丽艳, 王海亮, 刘智军. 乔木树种分布格局和林分空间结构分析[J]. 林业资源管理, 2020(1): 54-61. |
[4] | Masanori, O. and Takeshi, I. (2021) Explainable Identification and Mapping of Trees Using UAV RGB Image and Deep Learning. Scientific Reports, 11, 1-15. https://doi.org/10.1038/s41598-020-79653-9 |
[5] | 万意, 李长春, 赵旭辉, 刘冰洁. 基于SVM的光学遥感影像分类与评价[J]. 测绘地理信息, 2018, 43(6): 74-77. |
[6] | 赵雪清, 安晓东. 决策树与人工神经网络的对比分析[J]. 电脑开发与应用, 2007(11): 13-15. |
[7] | 林志玮, 丁启禄, 黄嘉航, 涂伟豪, 胡典, 刘金福. 基于DenseNet的无人机光学图像树种分类研究[J]. 遥感技术与应用, 2019, 34(4): 704-711. |
[8] | Geetha, G., Kirthigadevi, T., Ponsam, G.G., et al. (2020) Image Captioning Using Deep Convolutional Neural Networks (CNNs). Journal of Physics: Conference Series, 1712, Article ID: 012015.
https://doi.org/10.1088/1742-6596/1712/1/012015 |
[9] | 涂清芳. 琅琊山风景区野生观赏植物资源调查及评价[D]: [硕士学位论文]. 南京: 南京林业大学, 2012. |
[10] | 王欣, 樊彦国. 基于改进DenseNet联合空谱注意力机制的高光谱图像分类[J/OL]. 激光与光电子学进展, 1-19.
http://kns.cnki.net/kcms/detail/31.1690.tn.20210423.1116.018.html, 2021-10-10. |
[11] | 宋宇鹏, 边继龙, 安翔, 张锡英. 基于注意力机制的DenseNet模型的树种识别应用[J]. 实验室研究与探索, 2020, 39(7): 122-127+173. |
[12] | 吴云志, 刘翱宇, 朱小宁, 刘晨曦, 范国华, 乐毅, 张友华. FI-DenseNet: 用于植物病害图像识别的卷积网络[J/OL]. 安徽农业大学学报, 2021, 48(1): 1-7. https://doi.org/10.13610/j.cnki.1672-352x.20210319.002, 2021-10-10. |
[13] | 张玉红, 白韧祥, 孟凡军, 王思斯, 吴彪. 图像识别中的卷积神经网络应用研究[J]. 新技术新工艺, 2021(1): 52-55. |
[14] | Simonyan, K. and Zisserman, A. (2014) Very Deep Convolutional Networks for Large-Scale Image Recognition. |
[15] | Li, L.W., Qin, S.Y., Lu, Z., Xu, K.H. and Hu, Z.Y. (2020) One-Shot Learning Gesture Recognition Based on Joint Training of 3D ResNet and Memory Module. Multimedia Tools and Applications: An International Journal, 79, 5.
https://doi.org/10.1007/s11042-019-08429-9 |
[16] | 马金林, 张裕, 马自萍, 毛凯绩. 轻量化神经网络卷积设计研究进展[J/OL]. 计算机科学与探索, 1-21.
http://kns.cnki.net/kcms/detail/11.5602.tp.20210927.2250.004.html, 2021-10-10. |
[17] | 朱雪晨, 陈三林, 蔡刚, 黄志洪. 降低参数规模的卷积神经网络模型压缩方法[J]. 计算机与现代化, 2021(9): 83-89. |
[18] | 郭俊亮, 张洪川. 卷积神经网络模型研究分析[J]. 科技创新与应用, 2021, 11(23): 16-18+22. |