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

A Bidirectional Colorization Method for Anime Images Based on U-shaped Periodically Consistent Adversarial Networks

A colorization and network technology, applied in the field of image processing, can solve problems such as image quality loss, image data difficulties, compositional imbalance, etc., and achieve the effects of improving efficiency, strengthening generalization ability, and reducing workload

Inactive Publication Date: 2020-01-07
HEBEI UNIVERSITY OF SCIENCE AND TECHNOLOGY
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, for the animation sketch coloring problem, it is difficult to obtain paired image data
Cycle Consistent Adversarial Networks (CycleGAN), a new model that uses unpaired image-to-image datasets, still suffers from image quality loss and compositional misalignment

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
  • A Bidirectional Colorization Method for Anime Images Based on U-shaped Periodically Consistent Adversarial Networks
  • A Bidirectional Colorization Method for Anime Images Based on U-shaped Periodically Consistent Adversarial Networks
  • A Bidirectional Colorization Method for Anime Images Based on U-shaped Periodically Consistent Adversarial Networks

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0050] refer to Figure 1-Figure 4 , Embodiment 1 of the present invention relates to a two-way colorization method for animation images based on a U-shaped periodical consistent confrontation network, specifically as follows:

[0051] Step 1. Collect data: use crawlers to obtain high-definition full-color animation illustration images and black-and-white line draft images;

[0052] Step 2, setting the pixels of the animation illustration image to a uniform size, such as adjusting the pixels to 256×256, and putting it into a database to construct a training data set and a testing data set;

[0053] Step 3. Construct a U-shaped periodic consistent deep learning confrontation network, use the training data set obtained in step 2 to cyclically train the U-shaped periodic consistent deep learning confrontation network, and use the test data set to verify the performance of the U-shaped periodic consistent deep learning confrontation network;

[0054] Step 4. Input the test image in...

Embodiment 2

[0056] refer to Figure 1-Figure 4 , Embodiment 2 of the present invention relates to a two-way colorization method for animation images based on a U-shaped periodical consistent confrontation network. On the basis of the above-mentioned Embodiment 1, Embodiment 2 of the present invention is described in detail as follows:

[0057] The U-shaped periodic consistent deep learning confrontation network includes a generator G, a generator F, and a discriminant network D X and discriminative network D Y ; Among them, the generator G generates the input black-and-white line draft image into a full-color image, the generator F generates the full-color image into a black-and-white line draft image, and the discriminant network D X Judging whether the input black and white line draft image conforms to the distribution of the real black and white line draft image, the discriminant network D Y Determine whether the input full-color image conforms to the distribution of real full-color ...

Embodiment 3

[0059] refer to Figure 1-Figure 4 , Embodiment 3 of the present invention relates to a two-way colorization method for animation images based on a U-shaped periodical consistent confrontation network. On the basis of Embodiment 1 and / or Embodiment 2 above, Embodiment 3 of the present invention is described in detail as follows:

[0060] Improve the generator G, generator F, and discriminant network D through the method of U-shaped cycle consistent cycle training X and discriminative network D Y ability; generally speaking, the method of loop training in step 3, that is, by constructing generator G, generator F and discriminant network D Y The sub-recurrent network of G, training generator G and discriminative network D Y Ability; by constructing generator F, generator G and discriminant network D X The sub-recurrent network of , training generator F and discriminative network D X Ability.

[0061] In detail, the specific steps of the method for loop training in step 3 ar...

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 two-way colorization method for animation images based on a U-shaped periodic consistent confrontation network, relating to the Image processing field, Data collection, pixels of the animation illustration image are unified; A training data set and a test data set are constructed, finally, the capacities of a generator G, a generator F, a discriminating network DX and a discriminating network DY are improved by adopting a cyclic training method with consistent U-shaped periods, a function mapping relation between an image and a color image is found, and bidirectionalconversion of a black-white sketch and a full-color image is realized. According to the method, the features do not need to be extracted manually, a training set does not need to be marked, the workload of a cartoon creator is remarkably reduced, the image colorization and blackening processing efficiency is improved, and great help is provided for the cartoon creator.

Description

technical field [0001] The invention relates to a two-way colorization method of an animation image based on a U-shaped periodic consistent confrontation network, and belongs to the field of image processing. Background technique [0002] At present, the use of artificial intelligence for artistic creation is still in the testing stage. There are two models in the research area: [0003] (1) Neural Style Transfer is an image-to-image translation method that makes the content image have the style of the style image on the basis of Gram matrix statistics matching the pre-trained features. The point is that in Convolutional Neural Networks (CNN), content and style representations can be separated, and these two features can be operated independently to generate new perceptual images. However, neural type transfer is not up to the task. [0004] (2) Generative Adversarial Networks (GANs) are a new type of generative model and have achieved some results in image generation. Th...

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 Patents(China)
IPC IPC(8): G06T11/00G06T7/90
CPCG06T7/90G06T11/001G06T2207/20081G06T2207/20084
Inventor 张光华屈梦楠靳宇浩宋庆鹏吕月颖高永献栗彤张红斌
Owner HEBEI UNIVERSITY OF SCIENCE AND TECHNOLOGY
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