Underwater degraded image enhancement method based on GAN network

A degraded image and network technology, applied in the field of image processing and deep learning, can solve a large number of paired data, difficult to obtain and other problems, and achieve good visual effects

Pending Publication Date: 2020-06-09
BEIJING UNIV OF TECH
View PDF0 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

These methods significantly improve the visual effect and quality of underwater images, but require a large amount of paired

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
  • Underwater degraded image enhancement method based on GAN network
  • Underwater degraded image enhancement method based on GAN network
  • Underwater degraded image enhancement method based on GAN network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0019] 1. Build a data set

[0020] Training set: includes three types of X, Y and E, X is the underwater degraded image, Y is the underwater clear image, and E is the image in Y that has been Gaussian blurred. There is no one-to-one relationship between X-type images and Y-type images. Three types of images such as Figure 6a , b, c shown:

[0021] Test set: Contains only underwater degraded images, same as the X type in the training set.

[0022] 2. Model construction

[0023] The model consists of two modules: a generator module and a discriminator module. The generator module in this model contains two generators, G and F. The networks of the two generators are the same, and the parameters after training are different. The discriminator module contains only D X and D Y The two discriminators and the two generator networks are the same, but the parameters after training are different. The goal of the generator in the training process is to create fake samples so that...

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 relates to an underwater degraded image recovery method based on a GAN network. The method is used for solving the problem of underwater degraded image recovery without paired data sets.A network architecture of the method consists of two generator networks and two discriminator networks, two generators are respectively used for converting an underwater degraded image into a clear image and restoring the generated clear image into a degraded image, and the two generators have the same structure and different parameters. The two discriminators respectively determine whether an image generated by the first generator is a clear image and determine whether the image generated by the second generator is a degraded image. First, a training generator is arranged, when the number oftraining times is reached, generator parameters are not changed, a discriminator is trained again, when the discriminator reaches the number of training times, the discriminator parameters are not changed, the generator continues to be trained, the steps are repeated until the loss function is minimized to complete optimization training of the network, and the trained first generator is used forgenerating a clear image of the underwater degraded image.

Description

technical field [0001] The invention relates to the fields of image processing and deep learning, especially for enhancing tasks of underwater degraded images. Background technique [0002] Underwater degraded image enhancement task is one of the current research hotspots in the field of image processing. With the continuous development of marine resource development, underwater robots have become a high-tech means in the field of marine development and utilization. Underwater robots replace humans to complete tasks that humans cannot. Underwater robots mainly rely on visual capabilities. Therefore, in recent years, underwater image processing has become an important research direction. Under water, red light with a longer wavelength attenuates the fastest and disappears at 3-4 meters underwater. Blue light and green light with shorter wavelengths can travel farther in water. Therefore, videos collected by cameras in the sea or Images tend to appear blue-green. All kinds o...

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): G06T5/00
CPCG06T5/003G06T5/002G06T2207/20081G06T2207/20084
Inventor 胡永利王立国张勇王博岳员娇娇尹宝才
Owner BEIJING UNIV OF TECH
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
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap