In recent years,water depth inversion of satellite remote sensing images has been a hot research topic in China and abroad. Previous water depth inversion models of remote sensing images are mostly based on the condition of uniform sediment,lacking research on mixed seabed sediment. To address this issue,this paper proposed a water depth inversion model of remote sensing images based on a sediment classification. Satellite images of Wuzhizhou Island and its surrounding areas around Hainan Island in China were used as experimental data. After preprocessing and sediment classification,the bidirectional long short-term memory network (Bi-LSTM) model,Stumpf model,and one dimension-convolutional neural network (1D-CNN) model were used for water depth inversion. The water depth inversion results before and after sediment classification were analyzed,as well as the water depth inversion results of different models. The results show that the water depth inversion accuracy of different models after sediment classification is higher than that before sediment classification.The Bi-LSTM model has the highest water depth inversion accuracy,and the average absolute error,root mean square error,and determination coefficient of water depth inversion of remote sensing images after sediment classification are 0. 333 m,0. 474 m,and 0. 814 m,respectively,which are better than those of the comparison model.
维普
万方数据