Hyperspectral remote sensing image classification method based on self-attention context network

A technology of hyperspectral remote sensing and classification method, applied in the field of computer image processing, which can solve the problems of not fully considering network security and reliability, being vulnerable to the threat of adversarial attacks, and model prediction results deviating from the true labels of samples.

Active Publication Date: 2021-01-29
WUHAN UNIV
View PDF9 Cites 13 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, with the continuous deepening of research on adversarial attack algorithms in the field of computer vision [3], existing literature has found that existing deep neural networks are extremely vulnerable to adversarial samples, causing the model prediction results to deviate from the true labels of samples.
Considering that there is no research related to adversarial attacks in the field of hyperspectral remote sensing research, and the existing hyperspectral remote sensing image classification methods based on deep neural networks do not fully consider the security and reliability of the network during the design process, making these methods Highly vulnerable to adversarial attacks

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Hyperspectral remote sensing image classification method based on self-attention context network
  • Hyperspectral remote sensing image classification method based on self-attention context network
  • Hyperspectral remote sensing image classification method based on self-attention context network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0048] In order to facilitate those of ordinary skill in the art to understand and implement the present invention, the present invention will be described in further detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the implementation examples described here are only used to illustrate and explain the present invention, and are not intended to limit this invention.

[0049] The present invention provides a hyperspectral remote sensing image classification method based on a self-attention context network, which includes a backbone network, a self-attention module and a context coding module as an overall network. Among them, the backbone network extracts hierarchical features through three 3×3 dilated convolutional layers and a 2×2 average pooling layer. Subsequently, the features extracted by the backbone network are used as the input of the self-attention module to perform self-attention learning, and the spatial depende...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a hyperspectral remote sensing image classification method based on a self-attention context network. By self-attention learning and context coding, a spatial dependence relationship between pixels in a hyperspectral remote sensing image is constructed, and global context features are extracted. On the hyperspectral remote sensing data against attack pollution, the method can still keep excellent ground object recognition precision, so that the requirements of a hyperspectral remote sensing image classification task on safety and reliability are better met.

Description

technical field [0001] The invention belongs to the technical field of computer image processing, and relates to an image classification method, in particular to a hyperspectral remote sensing image classification method based on a self-attention context network. Background technique [0002] Hyperspectral remote sensing can simultaneously obtain continuous remote sensing observation data in spatial and spectral dimensions by combining spectral technology and imaging technology. Compared with natural images, hyperspectral remote sensing images have higher spectral resolution and more bands, which can reflect more abundant spectral characteristics of ground objects. Therefore, using hyperspectral images to classify and identify ground objects is one of the important ways to realize earth observation. [0003] At present, most hyperspectral image classification methods are based on deep convolutional neural networks [1-2], and have achieved good object recognition results. H...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/62G06K9/00G06N3/04
CPCG06V20/194G06V20/13G06N3/048G06N3/045G06F18/28G06F18/253G06F18/24G06V10/58G06V10/454G06V10/82G06V20/70Y02A40/10G06N3/0464G06N3/094G06N3/09G06V10/77G06V10/764
Inventor杜博徐永浩张良培
OwnerWUHAN UNIV