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

SAR image denoising method based on learning down-sampling and hopping connection network

A skip connection, image technology, applied in neural learning methods, image enhancement, image analysis, etc., can solve the problem that the spatial linear filtering method cannot completely preserve the edges and details, the speckle noise is not removed, the edge pseudo-Gibbs distortion, etc. problem, to reduce training loss, maintain image details, and expand the receptive field.

Inactive Publication Date: 2020-01-10
NANCHANG HANGKONG UNIVERSITY
View PDF2 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the characteristics of local processing, spatial linear filtering methods often cannot fully preserve edges and details
It has the following disadvantages: 1) the average value cannot be maintained, especially for the equivalent SAR image with a small visual range (ENL); 2) specific targets such as strong reflection points and small surface features are easily blurred or erased; 3) Speckle noise in dark scenes is not removed
However, it also has some disadvantages, such as poor translation robustness and pseudo-Gibbs distortion at the edges

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 denoising method based on learning down-sampling and hopping connection network
  • SAR image denoising method based on learning down-sampling and hopping connection network
  • SAR image denoising method based on learning down-sampling and hopping connection network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] The implementation of the present invention will be described in detail below with reference to the drawings and examples, so as to fully understand and implement the implementation process of how to use technical means to solve technical problems and achieve technical effects in the present invention.

[0036] like figure 1 As shown, a SAR image denoising method based on learning downsampling and skip connection network, including the following steps:

[0037] S1: Build an overall deep convolutional neural network model consisting of reversible downsampling, noise estimation and upsampling. The final output of the neural network is added to the source image to form a residual learning layer;

[0038] The reversible downsampling in the built overall deep convolutional neural network model reconstructs the input source image into four sub-images whose size is a quarter of the source image. As the input of CNN, it can effectively expand the receiving field, thereby improv...

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 denoising method based on a learning down-sampling and hopping connection network. Nonlinear end-to-end mapping between a noise image and a clean SAR image is realized by utilizing the learning down-sampling and hopping connection network (SAR-DSCN). The hopping connection network is added to the denoising model to maintain details of the image and reduce the disappearance gradient problem. And the receiving domain can be effectively expanded by adopting down-sampling. A large number of experiments on SAR images show that the method has better performance than the most advanced speckle suppression method at present, and the speed is higher than that of the traditional method. Results show that the effectiveness and high efficiency of SAR-DSCN enable theSAR-DSCN to have certain attraction in SAR image desquamation processing.

Description

technical field [0001] The invention relates to the technical field of SAR image denoising, in particular to a SAR image denoising method based on learning downsampling and skip connection network. Background technique [0002] Synthetic aperture radar (SAR) is an active coherent microwave radar that produces images with high spatial resolution. It has the characteristics of all-weather, day and night imaging, and high resolution. It has important application value in the field of remote sensing and plays an important role in military and civilian fields. However, under coherent radiation, the original image has a backscatter coefficient and uniform grain noise in the same area, that is, speckle noise. It is caused by constructive and destructive interference of coherent echoes scattered by smal reflectors within each resolution cell. The presence of speckle noise in SAR images often poses difficulties for processing and interpretation by computer vision systems and human ...

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): G06N3/04G06N3/08G06T5/00G06K9/62
CPCG06N3/08G06T2207/10044G06T2207/20081G06N3/045G06F18/214G06T5/70
Inventor 张向阳李仁昌邓召嵘高为民
Owner NANCHANG HANGKONG UNIVERSITY