Publications

Jing, WP; Lou, TT; Wang, ZY; Zou, WT; Xu, ZK; Mohaisen, L; Li, C; Wang, J (2023). A Rigorously-Incremental Spatiotemporal Data Fusion Method for Fusing Remote Sensing Images. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 16, 6723-6738.

Abstract
The spatiotemporal remote sensing images have significant importance in forest ecological monitoring, forest carbon management, and other related fields. Spatiotemporal data fusion technology of remote sensing images combines high spatiotemporal and high temporal resolution images to address the current limitation of single sensors in obtaining high spatiotemporal resolution. This technology has gained widespread attention in recent years. However, the current models still exhibit some shortcomings in dealing with land cover changes, such as poor clustering results, inaccurate incremental spatiotemporal calculations, and sensor differences. In this article, we propose a rigorously-incremental spatiotemporal data fusion method for fusing remote sensing images with different resolutions to address the aforementioned problems. The proposed method utilizes the particle swarm optimization Gaussian mixture model to extract endmembers and establishes a linear relationship between sensors to obtain accurate time increments. Furthermore, bicubic interpolation is used instead of thin plate spline interpolation for spatial interpolation, and also support vector regression is used to calculate weights for obtaining a weighted sum of temporal and spatial increments. In addition, sensor errors are allocated to the calculation of residuals. The experimental results show the efficacy of the proposed algorithm for fusing fine image Landsat with coarse image MODIS data and conclude that the proposed algorithm presents a better solution for heterogeneous data with strong phenological changes and regions with changes in surface types, which provides a better solution for remote sensing image fusion and, hence, improves the accuracy, stability, and robustness of data fusion.

DOI:
10.1109/JSTARS.2023.3296468

ISSN:
2151-1535