Unlock instant, AI-driven research and patent intelligence for your innovation.

A spatiotemporal fusion method of remote sensing images based on multi-scale mechanism and attention mechanism

A remote sensing image and space-time fusion technology, applied in the field of image processing, can solve the problems of low algorithm fusion accuracy and high data set requirements, and achieve the effect of alleviating the fusion results from being too smooth, high precision, and reducing the limitations of data sets

Active Publication Date: 2022-07-01
CHONGQING UNIV OF POSTS & TELECOMM
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Although there are various spatio-temporal fusion methods, there are still many problems. For example, in areas with high heterogeneity, the fusion accuracy of the algorithm is not high; The requirements are high; the fusion image obtained by the algorithm based on the convolutional neural network is usually too smooth

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
  • A spatiotemporal fusion method of remote sensing images based on multi-scale mechanism and attention mechanism
  • A spatiotemporal fusion method of remote sensing images based on multi-scale mechanism and attention mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0037]The technical solutions in the embodiments of the present invention will be described clearly and in detail below with reference to the accompanying drawings in the embodiments of the present invention. The described embodiments are only some of the embodiments of the invention.

[0038] The technical scheme that the present invention solves the above-mentioned technical problems is:

[0039] Specific steps:

[0040] Step S1, set t 0 and t 1 The high-temporal and low-spatial resolution images at the moment are used to make a difference, and the residual image M is obtained. 12 , which is fed into the first convolutional neural network, through the deconvolution layer to M 12 Perform upsampling to get M 12 Feature maps at three scales. will t 0 Low-temporal, high-spatial-resolution images of moments (L 1 ) input into another convolutional neural network, through the pooling layer to L 1 Perform downsampling to get L 1 Feature maps at three scales;

[0041] Step...

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 present invention claims to protect a remote sensing image spatiotemporal fusion method based on a multi-scale mechanism and an attention mechanism. Parallel convolutional neural network, extract their feature maps at different scales respectively; S2 selects feature maps on three scales, fuses them respectively, and then upsamples the fused feature maps on three scales to a unified scale, Then fuse them into a feature map; S3 inputs the fused feature map into the attention mechanism, and assigns different weights to the features and channels of the feature map; S4 combines the weights of the feature map and uses the fully connected layer to reconstruct the image , to obtain an image with high spatial and temporal resolution. The invention improves the accuracy of the remote sensing image spatiotemporal algorithm and has advantages in time efficiency.

Description

technical field [0001] The invention belongs to the technical field of image processing, and particularly relates to a spatiotemporal fusion method of remote sensing images based on a multi-scale mechanism and an attention mechanism. Background technique [0002] Limited by current hardware technology and budget constraints, the satellite images we obtain are inconsistent in temporal and spatial resolution. For example, the Moderate-resolution Imaging Spectroradiometer (MODIS) can acquire images of the same area every day or every half day. , but its spatial resolution is in the range of 250 meters to 1000 meters; while the United States' Landsat (Landsat) can obtain images with a spatial resolution of 10 meters to 30 meters, but its reentry period is 16 days. In practical applications, we need satellite imagery with high temporal and spatial resolution, so this problem can be solved by spatiotemporal fusion. The spatiotemporal fusion algorithm of remote sensing images is w...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G06T3/40G06N3/04G06N3/08
CPCG06T3/4046G06T3/4053G06N3/08G06N3/048G06N3/045
Inventor 李伟生张夏嫣
Owner CHONGQING UNIV OF POSTS & TELECOMM