Coastline detection algorithm based on MSDU-Net

Abstract

The invention belongs to the technical field of deep learning, and particularly relates to a coastline detection algorithm based on MSDU-Net, and the algorithm comprises the following steps: S1, data collection: collecting a coastline image data set, carrying out the manual marking of the edge of the coastline image data set, and completing the construction of a data set needed by model training; s2, data preprocessing: preprocessing comprises normalization and data division, a data set is amplified, and a model training effect is ensured; and S3, identifying the model: constructing a detection model by adopting a deep learning related technology, inputting training data, and completing the construction of a parameter model. And S4, model storage: when the loss function of the model is not reduced any more, storing the model. The coastline detection task is supported by fusing the texture features of different scales and the generated semantic features in the U-Net structure, and the shallow texture information and the deep semantic information are fused by combining the extractor with the U-Net network.

Description

technical field [0001] The invention belongs to the technical field of deep learning, and in particular relates to a coastline detection algorithm based on MSDU-Net. Background technique [0002] Unlike many other landmasses, Antarctica's coastline is surrounded by dynamic glaciers and fronts of ice shelves, constantly changing the position of the coastline through the disintegration of icebergs, which is affected by both seasonal changes and global climate change. Tracking the advancement and retreat of glaciers and ice shelf leading edges is an important factor in better understanding glacial processes. In addition, monitoring the retreat of the disintegration front is necessary because the disintegration front enhances the contribution of the Antarctic ice sheet to sea level due to weakened support. [0003] Problems or defects in the existing technology: In recent years, deep learning-based coastline detection algorithms have begun to surpass traditional statistical met...

AI Technical Summary

Problems solved by technology

However, they are often trained as single-purpose models for segmenting land and water or delineating coastlines, making it difficult to distinguish between open ocean and higher ice sheets

Images