Three-dimensional image restoration method based on deep learning

A technology of stereoscopic images and repair methods, applied in cross fields, can solve problems affecting 3D scene experience, inability to obtain information, affect filling effect, etc., and achieve the effect of improving application value

Pending Publication Date: 2020-02-07
BEIJING UNIV OF TECH
View PDF5 Cites 11 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In the case where the content in one view is lost but exists in another view, using a single image filling method will obviously affect the filling effect and 3D scene experience because the information of the relative view cannot be obtained.

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
  • Three-dimensional image restoration method based on deep learning
  • Three-dimensional image restoration method based on deep learning
  • Three-dimensional image restoration method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0027] Each module of the present invention, as well as the training strategy and the usage method of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0028] Module composition and flow process of the present invention are as figure 1 As shown, it specifically includes the following modules:

[0029] 1. Encoder and decoder modules

[0030] The Encoder and Decoder module consists of an encoder and a decoder. The encoder realizes feature extraction through the convolutional neural network, expands the receptive field to further feature condensing through the C-layer hole convolution on the basis of the same computing power and parameter quantity, and then obtains the feature expression of the semantic level through the D-layer convolutional neural network, and decodes The filter realizes the upsampling of the feature map through the deconvolution layer. In the example of the present invention, C=6,...

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 a three-dimensional image restoration method based on deep learning. The core of the method is a deep neural network model based on local adaptation and visual consistency. The model is composed of an encoder and decoder module, a fusion module, a discriminator module and a parallax module. According to the invention, for left and right views, the area of any size and anyposition can be restored, the application value is effectively improved, and the universality of image restoration is enhanced. Compared with the prior art, the method has the following advantages: 1)parts which can be referenced by each other are fused according to mutual complementation of the feature maps of the two views, so that not only can the features of the residual area of the view be referenced, but also the content of the other view can be referenced; meanwhile, the stereo consistency of the stereo image pair is constrained by using parallax clues, so that dizziness is not caused;and 2) the restored three-dimensional image pair is more vivid in texture and details, and is more in line with the visual perception of human eyes.

Description

technical field [0001] The invention belongs to the intersecting fields of digital image processing and computer vision, and relates to a stereoscopic image restoration method based on deep learning. Background technique [0002] In recent years, the rapid development of technology in the field of digital image processing and computer vision has driven the application of these technologies in other fields, and these applications have in turn prompted continuous technological innovation. Stereoscopic image inpainting is a subject that arises under such circumstances. Stereoscopic image repair is to repair the holes left on the stereoscopic image after removing the foreground or to complement the defects caused by the transmission process. This technology mainly uses the existing information of the image to repair the unknown area to be filled according to a certain algorithm according to clues such as color, texture, and parallax, so that the filled image has better effects ...

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): G06T5/00G06T5/50G06T15/00
CPCG06T5/001G06T5/50G06T15/00
Inventor 马伟郑玛娜
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