• Journal of Geo-information Science
  • Vol. 22, Issue 9, 1897 (2020)
Ang CHEN1, Xiuchun YANG1、2、*, Bin XU1、2, Yunxiang JIN1, Wenbo ZHANG1, Jian GUO1, Xiaoyu XING1, and Dong YANG1
Author Affiliations
  • 1Key Laboratory of Agri-informatics, Ministry of Agriculture/Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, China
  • 2College of Grassland Science, Beijing Forestry University, Beijing 100083, China
  • show less
    DOI: 10.12082/dqxxkx.2019.190598 Cite this Article
    Ang CHEN, Xiuchun YANG, Bin XU, Yunxiang JIN, Wenbo ZHANG, Jian GUO, Xiaoyu XING, Dong YANG. Research on Recognition Methods of Elm Sparse Forest based on Object-based Image Analysis and Deep Learning[J]. Journal of Geo-information Science, 2020, 22(9): 1897 Copy Citation Text show less
    Location map of study area in Hunshandake sandy land
    Fig. 1. Location map of study area in Hunshandake sandy land
    Flow chart of research methods
    Fig. 2. Flow chart of research methods
    Deep learning training samples
    Fig. 3. Deep learning training samples
    LV and LV-ROC Curves of UAV Image
    Fig. 4. LV and LV-ROC Curves of UAV Image
    Optimum segmentation of elm sparse forest in UAV image
    Fig. 5. Optimum segmentation of elm sparse forest in UAV image
    LV and LV-ROC Curves of GF-2 Image
    Fig. 6. LV and LV-ROC Curves of GF-2 Image
    Optimum segmentation of elm sparse forest in GF-2 image
    Fig. 7. Optimum segmentation of elm sparse forest in GF-2 image
    Sequencing of the Importance of UAV image features
    Fig. 8. Sequencing of the Importance of UAV image features
    Sequencing of the Importance of GF-2image features
    Fig. 9. Sequencing of the Importance of GF-2image features
    Accuracy comparison of SVM、RF、DNN used in OBIA
    Fig. 10. Accuracy comparison of SVM、RF、DNN used in OBIA
    Accuracy comparison between object-based method and deep learning method
    Fig. 11. Accuracy comparison between object-based method and deep learning method
    Comparison of extraction result between OBIA and deep learning for elm sparse forest
    Fig. 12. Comparison of extraction result between OBIA and deep learning for elm sparse forest
    飞行航高/m飞行海拔/m飞行速度/(km/h)航向重叠度/%旁向重叠度/%地面分辨率/m相片数/张测量面积/km2外扩面积/km2
    61019196075650.12452470.12
    Table 1. UAV flight parameters
    特征类型特征指标
    无人机GF-2
    指数特征VDVI、NGRDI、NGBDI、RGRI、EXGNDVI、MSAVI
    光谱特征Mean red、mean green、mean blue、mean H、mean L、mean S、std.dev.red、std.dev.green、std.dev.blue、std.dev.H、sed.dev.S、std.dev.L、brightness、max.diffMean red、mean green、mean blue、mean NIR、std.dev.red、std.dev.green、std.dev.blue、std.dev.NIR、brightness、max.diff
    冠层高度特征mean CHM
    纹理特征GLCM (Gray-Level Co-occurrence Matrix) homogeneity、GLCM contrast、GLCM dissimilarity、GLCM entropy、GLCM std. dev.、GLCM correlation、GLCM ang. 2nd moment、GLCM mean
    形状特征Area、number of pixels、border index、shape index、roundness、length/width
    Table 2. Feature list in OBIA
    类别背景地物榆树疏林总计
    背景地物2925317
    榆树疏林58325383
    总计350350700
    总体精度/%91.00
    Kappa0.82
    Table 3. Results of GF-2 image using deep learning method
    类别背景地物榆树疏林总计
    背景地物3489357
    榆树疏林2341343
    总计350350700
    总体精度/%98.43
    Kappa0.97
    Table 4. Results of UAV image using deep learning method
    Ang CHEN, Xiuchun YANG, Bin XU, Yunxiang JIN, Wenbo ZHANG, Jian GUO, Xiaoyu XING, Dong YANG. Research on Recognition Methods of Elm Sparse Forest based on Object-based Image Analysis and Deep Learning[J]. Journal of Geo-information Science, 2020, 22(9): 1897
    Download Citation