Sea ice classification method and system based on long-short-term memory network for SAR images

The invention belongs to the technical field of sea ice monitoring, a method and system for sea ice classification of SAR image based on long and short time memory network are disclosed, features areextracted directly from SAR image data of sea ice observed continuously as training input of long-term and short-term memory network, and sea ice concentration data are used as one-dimensional features to train the classification network in network training, and the sea ice classification network with both spatial and temporal dimensions is obtained. Considering the unknown sea ice density data, the sea ice density prediction model based on the long-short term memory network is trained first, and then the predicted density data and SAR image data set are input into the trained sea ice classification network to classify the sea ice. In the classification of sea ice in SAR image, the invention considers the time dimension characteristic of the sea ice type change in the time evolution process, and the recognition ra