Journal
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
Volume 61, Issue -, Pages -Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TGRS.2023.3307977
Keywords
Deep learning; feature fusion; implicit knowledge (IK); soil elements interpretation
Ask authors/readers for more resources
Soil elements are different types of soils with unique characteristics that are important for agriculture, ecological environment, and land permeability assessment. This study proposes a soil interpretation framework coupling implicit knowledge with a multispectral image (SIFCIM) to improve the accuracy of intelligent soil interpretation. Experimental results show that the proposed method outperforms interpretation methods with a single remote-sensing image, achieving significant improvements in overall pixel accuracy (oPA) and mean intersection over union (mIoU).
Soil elements refer to different types of soils with unique colors, textures, and particle sizes. Their interpretation is essential for agriculture, ecological environment, and land permeability assessment. This typically requires experts with dual knowledge of geology and remote sensing. With the increasing volume of remote-sensing data, the traditional visual interpretation and field survey technique is no longer sufficient to meet the demands. Because of the challenges such as the fine structure of soil, complex and variable natural scenes, and strong spatial variability, there remains a considerable gap between the accuracy of deep-learning-based methods and expert interpretation. To improve the accuracy of intelligent soil elements interpretation, this study proposes a soil interpretation framework coupling implicit knowledge with a multispectral image (SIFCIM). This framework quantifies implicit knowledge (IK), such as interpretation symbol and terrain feature, into matrix data [interpretation symbol distance (ISD) field and digital elevation model (DEM)]. To align with the SIFCIM, an IK-guided adaptive feature fusion network (IAFFNet) is constructed, which enhances the utilization efficiency of auxiliary features through an adaptive implicit feature fusion (AIF) module and a global feature dependence (GFD) module. Experimental results demonstrate that IAFFNet outperforms interpretation methods with a single remote-sensing image, achieving approximately 4.34% and 6.62% improvements in overall pixel accuracy (oPA) and mean intersection over union (mIoU), respectively. These results validate the effectiveness and robustness of the IK-guided approach in soil elements interpretation. To our knowledge, this work is the first to apply the concept of IK to soil elements interpretation, providing a novel insight for related research.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available