Satellite in-orbit application-oriented remote sensing image text intelligent description method

Abstract

The invention discloses a satellite in-orbit application-oriented remote sensing image text intelligent description method, which comprises the following steps of S100, data scaling and clipping: obtaining a remote sensing image for testing, and scaling and clipping the remote sensing image; s200, inputting into Encoder model for processing: importing trained model parameters, reusing the model toensure the effectiveness of the model, and outputting a feature map after the model passes through a multi-layer convolutional neural network; s300, inputting into a Decoder model for processing: inthe Decoder model based on the LSTM model, realizing feature mapping and word embedding of the image through the LSTM model; s400, generating text description: generating a remote sensing image semantic text description result under the constraint of an attention mechanism and a self-critical sequence training method in reinforcement learning; and S500, outputting a remote sensing image text description result. According to the invention, a residual network structure, two layers of LSTM models, an attention mechanism combining bottom to top and top to bottom and a self-critical sequence training method in reinforcement learning are introduced, and text semantic description can be rapidly and accurately generated for remote sensing images.

Description

technical field [0001] The invention belongs to the technical field of remote sensing, and in particular relates to a remote sensing image text intelligent description method for satellite in-orbit application. Background technique [0002] The rapid development of remote sensing technology provides a large amount of data accumulation for the acquisition of earth surface information, and at the same time promotes the rapid development of related technologies such as remote sensing image analysis and processing, establishment of high-resolution data sets, spatial data analysis, and network sharing. progress. Previously, the processing methods in the field of high-resolution remote sensing image interpretation have basically completed the transformation from pixel-oriented classification to object-oriented classification methods, but the interpretation level still stays at the object category level, and there is no reasoning and understanding of the scene. The "semantic gap" ...

AI Technical Summary

Problems solved by technology

Previously, the processing methods in the field of high-resolution remote sensing image interpretation have basically completed the transformation from pixel-oriented classification to object-oriented classification methods, but the interpretation level still stays at the object category level, and there is no reasoning and understanding of the scene. Unable to resolve the "semantic gap" between classification results and high-level scene semantics

Images