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

Single image rain line removal method based on deep convolutional neural network

A neural network and single image technology, applied in the direction of neural learning methods, biological neural network models, neural architecture, etc., can solve the problems of imaging blur, loss of detail information, image contrast reduction, etc., achieve parameter reduction, detail information elimination, The effect of feature map reduction

Inactive Publication Date: 2018-11-06
NANJING UNIV OF INFORMATION SCI & TECH
View PDF2 Cites 28 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Purpose of the invention: In order to overcome the deficiencies of the prior art, the present invention provides a single image rain line removal method based on a deep convolutional neural network, which can solve the problems of image contrast reduction, imaging blur, The problem of missing details

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 rain line removal method based on deep convolutional neural network
  • Single image rain line removal method based on deep convolutional neural network
  • Single image rain line removal method based on deep convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0031] Such as figure 1 As shown, the method shown in the present invention first uses guided filtering to decompose the rainy image into a low-frequency reference layer and a high-frequency detail layer. Then modify the objective function according to the domain knowledge of image processing, and input the high-frequency detail layer of the rainy image into the designed deep learning network architecture to learn the mapping between it and the high-frequency detail layer of the clear image corresponding to the rainy image. Finally, the high-frequency detail layer after the rain removed from the network is added to the low-frequency reference layer of the rain-bearing image to obtain a clear image after the rain line is removed. The present invention removes rain lines in a single image while retaining the details of the image after the rain is removed, so that the image definition is greatly improved.

[0032] (1) Perform image frequency division processing

[0033] First, the g...

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 rain line removal method based on the deep convolutional neural network. The method comprises steps that firstly, a rain image is decomposed into a low frequencyreference layer and a high frequency detail layer through utilizing guided filtering; secondly, a target function is modified according to the knowledge of the image processing domain, the high frequency detail layer of the rain image is inputted into the designed deep learning network architecture, and mapping between the high frequency detail layer of the rain image and the high frequency detail layer of a sharp image corresponding to the rain image is learned; and lastly, the rain-removed high frequency detail layer outputted by the network and the low frequency reference layer of the rainimage are superposed to obtain the rain-removed sharp image. The method is advantaged in that the detail part of the rain-removed image is remained when rain lines of a single image are removed, andthe image sharpness degree is greatly improved.

Description

Technical field [0001] The invention relates to a single image rainline removal method, in particular to a single image rainline removal method based on a deep convolution neural network. Background technique [0002] The research of single image rain removal is one of the important directions in the field of image restoration, which is widely used in object recognition, target tracking and other fields. However, a large number of fast-moving rain lines are randomly distributed in a rainy environment, so that reflection and refraction phenomena exist in the target object and the background light, resulting in reduced image contrast, blurred imaging, and loss of detailed information. The image is very difficult, so it is necessary to restore a single image with rain lines. [0003] The existing methods for removing rain lines from a single image are mainly divided into two categories. One class regards the problem as an image layer decomposition problem. It mainly includes struct...

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): G06T5/00G06N3/04G06N3/08
CPCG06N3/08G06T2207/20081G06N3/045G06T5/77
Inventor 郭业才李晨周腾威
Owner NANJING UNIV OF INFORMATION SCI & TECH
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