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

Neural network image defogging method based on hybrid convolution channel attention mechanism and hierarchical learning

A neural network and attention technology, applied in the field of image processing, can solve the problems of unsatisfactory restoration effect and difficult acquisition, and achieve the effect of improving the dehazing performance of the model

Pending Publication Date: 2020-08-14
WENZHOU UNIVERSITY
View PDF0 Cites 10 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, it is difficult to obtain accurate global atmospheric light and air scattering rate parameters in practical problems, which makes the restoration effect of the first method often unsatisfactory

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
  • Neural network image defogging method based on hybrid convolution channel attention mechanism and hierarchical learning
  • Neural network image defogging method based on hybrid convolution channel attention mechanism and hierarchical learning
  • Neural network image defogging method based on hybrid convolution channel attention mechanism and hierarchical learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0047] see Figure 1 to Figure 5 , a neural network image defogging method based on a hybrid convolution channel attention mechanism and layered learning disclosed by the present invention, comprising the following steps:

[0048] S1. Constructing an image dehazing model; wherein, the image dehazing model includes a multi-scale layered feature extractor, a mixed convolution channel attention module, and an image reconstruction module;

[0049] The specific process is as figure 2 As shown, the image dehazing model is constructed. Image dehazing models include multi-scale hierarchical feature extractors (such as figure 2 shown), hybrid convolutional channel attention modules (such as figure 2 shown) and image reconstruction module (such as figure 2 shown).

[0050] S2. Acquire foggy image data, and use the above-mentioned multi-scale layered extractor to extract feature maps of six different scales and different depths of the fog image in stages;

[0051]The specific p...

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 neural network image defogging method based on a hybrid convolution channel attention mechanism and hierarchical learning. The method comprises the following steps: constructing an image defogging model; acquiring foggy image data, and extracting feature maps of six different scales and different depths of a foggy image by stages by using a multi-scale layered extractor;constructing a hybrid convolution channel attention module based on hybrid convolution and an attention mechanism, and processing the fused features of the six feature maps by the hybrid convolution channel attention module to make a defogging model pay attention to effective features and perform feature defogging; reconstructing the defogged features into a clear fog-free image through an image reconstruction module; and calculating the loss of the restored image and the clear image thereof, and optimizing the image defogging model. According to the technical scheme, effective defogging processing is carried out on the actually shot fog image, and a high-quality fog-free image is recovered.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a neural network image defogging method based on a mixed convolution channel attention mechanism and layered learning. Background technique [0002] In recent years, in outdoor traffic monitoring and other fields, the camera system is often affected by harsh environmental weather such as rain, snow, fog, haze, etc. The pictures captured by the camera system affected by the environment will affect the normal work of monitoring personnel or tracking applications , therefore, it is of great significance to restore the image degraded by the influence of haze and so on. [0003] Image dehazing is one of the key issues in image restoration. The fog map can be modeled using the atmospheric light scattering model. The model is as follows: [0004] I=tJ+A(1-t) [0005] t(x)=e βd(x) [0006] Among them, I is the foggy image, t is the air scattering rate, J is the underlying cl...

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
IPC IPC(8): G06T5/00G06N3/04G06N3/08
CPCG06N3/08G06T2207/20221G06T2207/20081G06T2207/20084G06N3/045G06T5/73Y02T10/40
Inventor 张笑钦王涛王金鑫赵丽
Owner WENZHOU UNIVERSITY
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