Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Multi-temporal SAR image change detection method based on deep learning

An image change detection and deep learning technology, applied in the field of image detection, can solve the problem of introducing uncertainty in classification, and achieve the effect of suppressing speckle noise, reducing uncertainty, and suppressing human interference

Active Publication Date: 2020-04-28
CHONGQING UNIV
View PDF8 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this pixel-centric rectangular patch processing introduces artifacts on the boundary of the classification patch, which often introduces uncertainty in the classification

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
  • Multi-temporal SAR image change detection method based on deep learning
  • Multi-temporal SAR image change detection method based on deep learning
  • Multi-temporal SAR image change detection method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0057] The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0058] Such as figure 1 As shown, the present invention discloses a multi-temporal SAR image change detection method based on deep learning, including:

[0059] S1. Obtain the SAR image I of the first moment of detection target 1 and the SAR image I at the second moment 2 , I 1 and I 2 The dimensions are M×N;

[0060] S2, to I 1 and I 2 Perform superpixel segmentation to get I 1 and I 2 superpixel block, I 1 and I 2 The corresponding superpixel blocks of are equal;

[0061] Equal superpixel blocks mean that the corresponding superpixel blocks have the same shape, corresponding position, and number of pixels contained in the block. Because we first 1 Perform superpixel segmentation, and then use I 1 The split mode to split the I 2 .

[0062] S3. Reshaping the superpixel block to obtain a superpixel vector;

[0063] S4. Generate a superpixel ve...

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 multi-temporal SAR image change detection method based on deep learning. Compared with the prior art, neighborhood information is substituted into classification and clustering is carried out by taking super pixels as units, so that human interference during rectangular patch generation is inhibited, clustering uncertainty is reduced, and speckle noise influencing interpretability of the SAR image is also inhibited. Moreover, change detection is used as classification of two stages, so that a large number of false alarms caused by speckle noise are suppressed. In a first stage, we simply aggregate DI into varying and non-varying classes. In the second stage, low-rank sparse decomposition (LRSD) is adopted for preprocessing based on the internal difference betweenthe change caused by speckle noise and the change of the real object. The low-rank term of the LRSD enables the false change caused by the speckle noise to be recovered to the original state, and thesparse term separates the speckle noise from the image, thereby greatly weakening the impact on the subsequent classification from the speckle noise.

Description

technical field [0001] The present invention relates to the field of image detection, in particular to a multi-temporal SAR image change detection method based on deep learning. Background technique [0002] In the past few decades, synthetic aperture radar (SAR) images have attracted extensive attention in the fields of military affairs, environmental monitoring, and urban planning due to the fact that they are not limited by time and weather. One common use of this is change detection. Given two SAR images acquired from the same observation area at different times, the purpose of change detection is to identify the differences between them. We classify change detection methods into two categories according to whether there is a difference image DI (difference image). Post-classification comparison is to directly analyze the changed and unchanged regions in the two images that have been independently classified before analysis, thus avoiding the influence of radiation nor...

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 Applications(China)
IPC IPC(8): G06T7/246G06T7/11G06K9/62G06K9/00
CPCG06T7/246G06T7/11G06T2207/10044G06T2207/20081G06T2207/20084G06V20/13G06F18/24G06F18/214
Inventor 张新征刘过苏杭李道通周喜川
Owner CHONGQING UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com