A bathymetric inversion method based on adaptive weight fusion of multiphase remote sensing images

Remote sensing bathymetric inversion without in-situ data can provide important shallow-sea baseline data for navigation, island development and ecological restoration in the absence of measured data. To suppress the noise of single-phase images and overcome the problem of insufficient fusion of traditional multiphase images on the accuracy of bathymetric inversion, this paper proposes a bathymetric inversion method based on adaptive weight fusion of multiphase remote sensing images. The proposed method supplements the inversion results of single-phase by fitting deep-sea (the normalized difference water index below 0.05) data gaps. Based on the distance between the bathymetry of the homonymous points and average depth, the fusion weights of the homonymous points are adaptively calculated. The bathymetry of each homonymous point is fused to generate an accurate bathymetric result based on the adaptive weight fusion. Three different types of shallow-sea areas were selected for the experiments. The results show that the proposed method effectively suppresses noise in single-phase images and enhances inversion accuracy overall and across various bathymetric intervals. Compared with median fusion and average fusion, the inversion accuracy of the proposed method is significantly improved, more comprehensive bathymetric features are captured, and optimal bathymetric inversion results are achieved in different types of study areas.