Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Hyperspectral image deep noise reduction method and system based on two-stage learning framework

A hyperspectral image, two-stage technology, applied in the field of hyperspectral image noise reduction, can solve problems such as affecting the effect of image application and image distortion, and achieve the effect of solving image distortion and realizing deep noise reduction.

Pending Publication Date: 2022-06-28
XI AN JIAOTONG UNIV
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problems in the prior art, and provide a method and system for deep noise reduction of hyperspectral images based on a two-stage learning framework, aiming to solve the problem of mining HSI global space and spectral related information in the prior art , resulting in image distortion and defective technical problems that affect the effect of subsequent application of images

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 image deep noise reduction method and system based on two-stage learning framework
  • Hyperspectral image deep noise reduction method and system based on two-stage learning framework
  • Hyperspectral image deep noise reduction method and system based on two-stage learning framework

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0154] 1. HSI denoising experiment based on zero mean noise

[0155] First add zero-mean Gaussian noise to the CAVE, ICVL and Washington DC Mall datasets with noise levels 30, 50, 70 and random IID Gaussian noise. The quantitative evaluation results of various denoising methods on the CAVE, ICVE, and Washington DC datasets in the case of zero-mean Gaussian noise are shown in Table 1. It is not difficult to see that the proposed 3D-DUSSD model achieves better performance than other methods on most evaluation metrics. In addition, FastHyDe and HSI-SDeCNN and QRNN3D both show excellent denoising performance, but FastHyDe does not work well for blind noise, HSI-SDeCNN is good at knowing the noise level in advance, QRNN3D uses different noise intensities samples to train the network, but its model needs to be fine-tuned for a different dataset. The method uses only non-IID Gaussian noise to simulate training samples to train a single model, and shows better performance on differe...

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 image depth noise reduction method and system based on a two-stage learning framework, and belongs to the technical field of hyperspectral image noise reduction, and a three-dimensional depth universal model 3D-DUSSD for hyperspectral image noise reduction comprises the following steps: constructing a target function based on noise estimation and image noise reduction; constructing a conditional estimation sub-network (CENet) and a multi-scale cross fusion noise reduction sub-network based on a target function; training a conditional estimation sub-network (CENet) and a multi-scale cross fusion noise reduction sub-network by using an objective function; the trained conditional estimation sub-network (CENet) is utilized to deduce the noise level of the hyperspectral image, and then information is transmitted into the multi-scale cross fusion noise reduction sub-network to carry out the denoising of the hyperspectral image. By means of the method, non-i. I. D. Noise distribution can be predicted, and meanwhile deep noise reduction of the hyperspectral image is achieved.

Description

technical field [0001] The invention belongs to the technical field of hyperspectral image noise reduction, and relates to a method and system for deep noise reduction of hyperspectral images based on a two-stage learning framework. Background technique [0002] Hyperspectral data provides continuous or discontinuous 10nm bands in the 400–2500nm region of the electromagnetic spectrum. It is three-dimensional (3D) data composed of two-dimensional (2D) spatial information and one-dimensional (1D) spectral information. It has been widely used in agriculture, mineral identification, military surveillance and other fields. However, HSI is often affected by various factors in the process of acquisition and transmission, and is inevitably polluted by various noises including Gaussian noise, deadline noise, stripe noise, impulse noise, etc., which seriously affects the subsequent HSI classification. , target recognition, etc. Therefore, HSI denoising is a very important image prep...

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): G06T5/00G06N3/04G06N3/08
CPCG06N3/08G06T2207/10032G06T2207/20081G06T2207/20084G06N3/045G06T5/70
Inventor 刘帅许翔陈泽山肖嘉华高宗昂
Owner XI AN JIAOTONG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Eureka Blog
Learn More
PatSnap group products