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

Hybrid degraded image enhancement method based on convolutional neural network

A convolutional neural network and degraded image technology, applied in the field of computer vision enhancement, to achieve the effect of resolution enhancement, denoising enhancement, and good enhancement effect

Active Publication Date: 2021-05-14
WUHAN UNIV
View PDF5 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0009] At the problem of existing research method, the present invention research content comprises the following several parts:

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
  • Hybrid degraded image enhancement method based on convolutional neural network
  • Hybrid degraded image enhancement method based on convolutional neural network
  • Hybrid degraded image enhancement method based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035]The present invention can use a computer to train and infer the network, and use the Tensorflow depth learning framework under the Windows operating system. The specific experimental environment is configured as follows:

[0036] platform Google Colaboratory & Google Drive processor Intel (r) Xeon (r) cpu@2.30GHz x2 GPU NVIDIA TESLA P100-PCIE operating system Ubuntu 18.04.3lts Programming language Python 3.6.9 Deep learning framework Tensorflow 1.13, Keras 2.2.5

[0037]The specific implementation is as follows:

[0038]Step 1: Data set expansion. The training data set for the experiment is the image super-resolution reconstruction training set BSDS200, which contains 200 321x481-sized PNG format images, including characters, animals, plants, buildings, and various natural landscapes. The test dataset is an image super-resolution reconstruction test set set5. Before training, the image data is pre-processed in order to expand the data set and simulating the degraded image.

[003...

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

In order to solve the problem that a single image enhancement model cannot process a mixed degraded image, the invention provides a hybrid degraded image enhancement method based on a convolutional neural network, namely a fully convolutional sub-pixel residual dense network (FCSRDN), the network is based on the fully convolutional neural network (FCN), the residual network and the dense network are connected through a ResNet and a dense Net, and the enhanced model of the mixed degraded image based on the FCSRDN is used for enhancing the mixed degraded image based on the FCSRDN. The method comprises the following steps of: carrying out feature fusion on an original image, finishing feature fusion of each level of the original image, carrying out up-sampling by combining a sub-pixel convolutional layer, and enlarging the size of the image to finally obtain an enhanced image. According to the method, network input of any size can be accepted, compared with an existing method, enhancement in the three aspects of contrast ratio improvement, resolution ratio improvement, denoising and the like can be completed at the same time, and good enhancement effects are achieved in all aspects.

Description

Technical field[0001]The present invention belongs to a computer visual enhancement technique, and more particularly to an image enhancement model based on a convolutional neural network of a mixed degraded image.Background technique[0002]Image is one of the most important media of information propagation in the current information, and there is a pivotal position in numerous applications. However, due to the problem of imaging equipment, imaging environments, photographing personnel technical levels and other problems, a large number of images exist such as degradation of low brightness, low brightness, noise, color matrix, distortion, and other degradations. The purpose of image enhancement is to enhance useful information in the image, enhance image quality, improve image visual effects.[0003]Traditional image enhancement technology, after a long period of development, technology is gradually sure, and has achieved good performance in many fields. Traditional image enhancements c...

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): G06T5/00G06T3/40G06N3/04G06N3/08
CPCG06T3/4046G06N3/08G06T2207/20081G06T2207/20084G06N3/045G06T5/90G06T5/70
Inventor 曹丽琴宋争光金佳惠李治江
Owner WUHAN UNIV
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