Unlock instant, AI-driven research and patent intelligence for your innovation.

Single-image super-resolution reconstruction method based on deep learning

A super-resolution reconstruction, single image technology, applied in the field of computer vision, can solve the problem of difficulty in recovering high-frequency knowledge, and achieve the effect of improving the peak signal-to-noise ratio, saving computing resources, and enhancing transmission.

Active Publication Date: 2019-11-19
NANJING UNIV OF SCI & TECH
View PDF2 Cites 12 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, low-resolution images contain mostly low-frequency knowledge, and high-frequency knowledge is difficult to recover. Previously, most of the GAN networks recovered were high-frequency noise rather than high-frequency information.

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
  • Single-image super-resolution reconstruction method based on deep learning
  • Single-image super-resolution reconstruction method based on deep learning
  • Single-image super-resolution reconstruction method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0078] In order to verify the effectiveness of the scheme of the present invention, this example sets the magnification factor to 4, and performs comparative experiments on three standard image test sets Set5, Set14, and BSD100. The super-resolution reconstruction results are as follows Figure 1-3 The objective evaluation indicators are shown in Table 1.

[0079] By comparing the images generated by the algorithm of the present invention with those generated by Bicubic, SelfEx, and SRCNN, it can be intuitively found that the super-resolution results of other methods lack high-frequency information, and the images tend to be blurred, but the algorithm of the present invention can better restore the High-frequency information such as texture details, the image is also clearer, which has obvious advantages in intuitive experience. Such as figure 1 As shown, the baby's eyebrows, eyelashes, the pattern of butterfly wings, and the hair on the side of the face can be well restored....

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 single image super-resolution reconstruction method based on deep learning, which removes a batch normalization layer of a residual module for a network architecture, adds anetwork which is tightly connected and acts on a discrimination domain, and fuses a plurality of different loss functions for a loss function. Computing resources are saved. communication between layers is enhanced. The generated image has high-frequency information rather than high-frequency noise, and the peak signal-to-noise ratio, the structural similarity and the visual effect of the generated image on different data sets are improved.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a single image super-resolution reconstruction method based on deep learning. Background technique [0002] Single image super-resolution is an important branch of computer vision. It aims to generate a corresponding high-resolution image from a low-resolution image through a convolutional neural network. It is useful in pedestrian detection, vehicle detection, face recognition and other scenarios. Wide range of applications. At present, the key problem to be solved in super-resolution is the restoration of high-frequency texture details. The key to inferring high-definition images from a low-resolution image is how to obtain high-frequency knowledge such as edge textures. However, low-resolution images contain mostly low-frequency knowledge, and high-frequency knowledge is difficult to recover. Most of the GAN networks recovered before were high-frequency noise rather than high-f...

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): G06T3/40G06K9/62G06N3/04
CPCG06T3/4053G06T3/4023G06T2207/20081G06T2207/20084G06N3/045G06F18/214
Inventor 杜天文张毅锋束锋刘林桂林卿张一晋
Owner NANJING UNIV OF SCI & TECH