Remote sensing image natural language generation method based on attention mechanism and deep learning

Abstract

The invention relates to a remote sensing image natural language generation method based on an attention mechanism and deep learning. The method comprises the steps of first, preprocessing a remote sensing image and corresponding natural language description; second, inputting the denoised remote sensing image into an intensive positioning convolution neural network (IPCNN); third, inputting region blocks obtained in the second step into a reassignment long-short term memory (RLSTM) network, getting into a weight assignment layer of the RLSTM network, respectively solving the weight of each region by using a multilayer network function, and finally realizing overall output of the natural language description through a deep output layer of the RLSTM network; fourth, inputting the natural language description generated in the third step into a remote sensing image language description scoring model to obtain a score of each sentence; and fifth, inputting a target position, a category label and a natural language description score into a database to wait for calling.

Description

technical field [0001] The invention relates to attention mechanism, deep learning, and natural language generation of remote sensing images. Background technique [0002] In recent years, aerospace technology, as the core technology for obtaining information, has developed very rapidly. As one of the important branches, satellite remote sensing technology has not only greatly improved the resolution, but also gradually possessed the ability of all-day, all-weather, and real-time transmission . As an effective carrier of information, high-resolution remote sensing images broaden the field of vision of the human eye, improve the accuracy of target observation, and play a very important role in resource exploration, environmental monitoring, and natural disaster prevention. [0003] With the increasing demand for remote sensing applications, how to obtain information in remote sensing images conveniently and quickly is an important research direction. Especially general user...

AI Technical Summary

Problems solved by technology

[0006] The current natural language description methods for remote sensing images have certain limitations: (1) Remote sensing images generally have large pixels and small proportion of objects

In the training phase of the convolutional neural network, the general solution mode is to directly input the original image and its annotation information into the network, which causes the network to ignore a lot of background information and only focus on the label of the target feature information.

(2) Since the traditional natural language processing adopts the classical paradigm framework, this framework is designed based on the template method and has language limitations

The given description is modeled, not flexible and user-friendly, and a lot of information is omitted in the model, resulting in the loss of information

Images