Check patentability & draft patents in minutes with Patsnap Eureka AI!

SAR image road segmentation method based on attention mechanism

An attention and attention model technology, applied in the field of image processing, can solve the problems of unintuitive SAR image representation, difficult processing, and the difference between the target and the background is not obvious, so as to improve the performance of target segmentation and reduce speckle interference. Influence, avoid the effect of missed detection and false detection

Pending Publication Date: 2021-06-01
SHAANXI NORMAL UNIV
View PDF0 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Because the SAR image representation is not intuitive, there are spots, and the difference between the target and the background is not obvious, it is difficult to process it, so there are few studies on the use of deep learning to segment roads in SAR images, and it is currently used for The neural network involved in the method of this direction was proposed earlier, and its segmentation accuracy and convergence speed are insufficient.

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
  • SAR image road segmentation method based on attention mechanism
  • SAR image road segmentation method based on attention mechanism
  • SAR image road segmentation method based on attention mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0043] The SAR road data used in steps S2 and S3 in this embodiment of the present invention comes from 23 SAR images of Gaofen-3 in Shaanxi, China, and consists of 10,026 road samples with a size of 512*512 pixels, of which 70% of the training data set is used for verification. The dataset accounts for 20%, and the test dataset accounts for 10%. The imaging modes of the images in the dataset include four types: spotlight, hyperfine strip, fine strip 1, and fine strip 2, with resolutions covering 1m, 3m, 5m, and 10m. In addition, the road shapes in the data set include three-fork roads, cross roads, winding roads, etc., and the road backgrounds include farmland, villages, towns, etc., which can effectively avoid the over-fitting phenomenon of deep learning algorithms in road segmentation to a certain extent.

[0044] The construction process of the dataset is as follows Figure 10 shown

[0045]1) The size of the original 23-scene GF-3 SAR image is about 13200*24300. Select...

experiment example

[0086] In order to verify the segmentation effect of the method of the present invention on SAR image roads, we selected 3500 marked 512*512 pixel SAR images as the training set, and divided them into three batches for training, 500 images, 1000 images and 2000 images respectively. Zhang, taking the training of 1000 images as an example, observed the change trend of the loss value of the model, and found that the model can achieve convergence faster. At the same time, we select 4 pictures as test images, the original test images and the test results based on the traditional Mask RCNN algorithm and the algorithm proposed by the present invention are as attached Figure 7 , 8, 9, it can be seen that the robustness of the method of the present invention is strong, and the method can still segment the road target well under the influence of spots, and the segmentation accuracy is very high. The specific contour is easy to observe, and there is no need for artificial secondary pro...

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 an SAR image road segmentation method based on an attention mechanism, and relates to the technical field of image processing. The method comprises the steps: constructing a segmentation network model comprising a convolution block attention model and a Mask RCNN network; S2, training the segmentation network model constructed in the step S1 by using an SAR road data set; and inputting the SAR image to be segmented into the trained segmentation network model to segment the road. According to the SAR image road segmentation method, the defects in the prior art are overcome, the target segmentation performance is improved, more useful information is extracted from the source image, the influence of spot interference is reduced, and the situations of missing detection, false detection and the like are avoided.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a SAR image road segmentation method based on an attention mechanism. Background technique [0002] Synthetic Aperture Radar (SAR) is a high-resolution coherent imaging radar, which not only has the ability to work all-day and all-weather, but also has abundant characteristic signals, including amplitude, phase and polarization and other information. Therefore, road segmentation from SAR images has received increasing attention. The SAR imaging mechanism is complex, and it is easy to form multiplicative coherent speckle noise, which makes the additive noise model edge detection operator suitable for optical remote sensing images no longer applicable in SAR images, and the existence of speckles also seriously affects the performance of SAR images. interpret. [0003] In recent years, according to the characteristics of SAR images, many methods have been proposed, such a...

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): G06K9/00G06K9/32G06K9/34G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/084G06V20/588G06V10/25G06V10/267G06V10/44G06N3/048G06N3/045G06F18/253
Inventor 孙增国耿惠陈昱莅刘明吴迪
Owner SHAANXI NORMAL UNIV
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com