Author(s): Xiang Daxiang; Li Jingwei; Wu Yibang; Chen Zhe; Li Zhe; Jiang Ying
Linked Author(s):
Keywords: Multi-source remote sensing; Feature extraction; Geometric registration; Soil erosion
Abstract: Using high-resolution remote sensing images to calculate the loss of water and soil can provide scientific and objective data basis for soil and water conservation monitoring and supervision. However, multi-source remote sensing image registration errors often cause deviations in the calculation of soil erosion. To solve this problem, This paper constructs an improved SIFT operator based on both scale factor weighted constraints and Hellinger transform to extract multi-source image matching features. A model-consistent region division method combined with hierarchical clustering is used to analyze the degree of compliance of feature information with different transformation models in the image. The characteristics of image are then calculated by region, and local models are applied to complete the registration from coarse to fine. Experiments show that the registration accuracy of multi-source remote sensing images has a great impact on the accuracy of calculation of soil erosion. The average change rate of the loss of water and soil before and after the improved registration method is 16.84%, especially in the boundary areas of different surface types. This method can effectively improve the calculation accuracy of soil erosion.
Year: 2024