An image complement method based on a generated antagonistic network model

A network model and image technology, applied in the field of deep learning neural network, can solve the problems of slow training speed, no automatic completion of images, etc., and achieve the effect of high efficiency

Inactive Publication Date: 2019-02-19
SOUTH CHINA UNIV OF TECH
View PDF5 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In this case, the whole network is slower to

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
  • An image complement method based on a generated antagonistic network model
  • An image complement method based on a generated antagonistic network model
  • An image complement method based on a generated antagonistic network model

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0025] Example

[0026] This embodiment discloses an image completion method based on generating a confrontation network model, which specifically includes the following steps:

[0027] Step S1: Construct an original generative confrontation network model, and the generator inputs the generated image to the discriminator for network training.

[0028] Step S2: Construct a deep convolutional neural network as a generator and discriminator;

[0029] Different convolution kernels are reflected in the different matrix values ​​and the number of rows and columns.

[0030] Construct multiple convolution kernels. In the process of image processing, different convolution kernels means that different features of the generated image can be learned during the network training process.

[0031] In the traditional confrontation network model, the discriminator receives random noise, and by continuously learning the distribution of the data set, the random noise is generated to meet the distribution o...

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 image complement method based on a generated antagonistic network model, and belongs to the field of depth learning neural network, comprising the following steps: S1, constructing an original generated antagonistic network model; 2, constructing a depth convolution neural network as a generator and a discriminator; S3, removing part of pixels in the data set image and inputting them into the generator; S4, using convolution neural network to complete the image in the generator; S5, inputting the completed image and the dataset image into a discriminator to discriminate, and updating the loss function. In this method, the generated antagonistic network model based on image complement is constructed, which changes the information received by the generator from noise to image with partial pixels removed. Through antagonistic training between the generator and the discriminator, the generator can automatically complete the missing partial pixels.

Description

technical field [0001] The invention relates to the technical field of deep learning neural network, in particular to an image completion method based on a generative confrontation network model. Background technique [0002] Generative Adversarial Network (GAN for short) is a deep learning framework proposed by Goodfellow in 2014. It is based on the idea of ​​"game theory" and constructs two models, the generator and the discriminator. The former The image is generated by inputting (0, 1) uniform noise or Gaussian random noise, which discriminates the input image to determine whether it is an image from the dataset or an image produced by the generator. [0003] In the traditional confrontational network model, the discriminator receives random noise, and by continuously learning the distribution in the data set, the random noise is generated into an image that satisfies the distribution of the data set. In this case, the whole network is slow to train and there is no auto...

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
CPCG06T5/005G06T2207/20081G06T2207/20084
Inventor 周智恒李立军
Owner SOUTH CHINA 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