Residual single image rain removal method based on attention mechanism

An attention and residual technology, applied in the field of image processing, can solve the problems that the convolutional neural network structure is too simple and cannot achieve the removal effect, and achieve the effect of simple network training process, good rain removal effect, and reduced interference

Active Publication Date: 2021-06-04
XIDIAN UNIV
View PDF5 Cites 12 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that the designed convolutional neural network structure is too simple and cannot achieve a good removal effect.

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
  • Residual single image rain removal method based on attention mechanism
  • Residual single image rain removal method based on attention mechanism
  • Residual single image rain removal method based on attention mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0031] Refer to attached figure 1 , a method for removing rain from a residual single image based on an attention mechanism proposed by the present invention, comprising the following steps:

[0032] Step 1: Preprocess the input image to obtain a rainy image after preprocessing;

[0033] The preprocessing is specifically: normalize the pixel value of the input image to [0, 1], and crop the pixel size to 256*256*3.

[0034] Step 2: Build an attention residual neural network model;

[0035] (2.1) Convolutional residual blocks composed of convolutional layers, short skip connections, and long skip connections are used to form a residual network module for image feature extraction of rainy images after preprocessing;

[0036] Such as figure 2 As shown, in this embodiment, one convolutional layer, five convolutional residual blocks composed of short skip connections, and one long skip connection constitute a residual network module, which is used to extract image features from ...

Embodiment 2

[0058] refer to Figure 4 , a schematic diagram of a deep learning model of the method of the present invention;

[0059] The present invention's method for removing rain from a residual single image based on the attention mechanism is the same as in Embodiment 1, wherein the attention residual neural network model constructed in step 2, the image processing process is specifically as follows:

[0060] 4.1, input a rainy image, after the first convolutional layer processing, the output feature F1, the number of channels is 32, the output feature F1 is processed by 5 convolutional residual blocks composed of short skip connections, and the output feature F2, the number of channels is 32;

[0061] 4.2, the output feature F1 and the output feature F2 are connected by long jumps, and the output feature F3 is output;

[0062] 4.3, the output feature F3 is processed by the channel attention module SE_Block, and the output feature F4;

[0063] 4.4, the output feature F4 undergoes ...

Embodiment 3

[0082] The residual single image deraining method based on attention mechanism is the same as embodiment 1-2, the channel attention module SE_Block described in step (2.2.1), such as Figure 5 As shown; the specific steps for image features to be processed by channel attention module SE_Block include:

[0083] 5.1, after the image features of the input image pass through the global pooling layer, fully connected layer, ReLU activation layer, fully connected layer and Sigmoid layer in turn, output the weight of each channel;

[0084] 5.2. Multiply the weight of each channel output in step 5.1 with the input image feature of step 5.1 channel by channel, and output the processed image feature.

[0085] The present invention uses the residual network and codec network structure to extract more detailed image features, fully obtains the rain pattern information of the rainy image, and adopts the attention mechanism to further obtain more detailed information of the rain pattern in ...

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 residual single image rain removal method based on an attention mechanism, and mainly solves the problems that the existing single image rain removal technology has limitation and is not ideal in processing effect. According to the scheme, the method comprises the following steps: 1) preprocessing an input image to obtain a preprocessed image; 2) constructing an attention residual neural network model comprising a residual network module and a codec network module; 3) inputting the preprocessed image into an attention residual neural network model for training, constraining the attention residual neural network model by using a loss function, and then performing back propagation for parameter updating to obtain a trained rain removal neural network model; and 4) inputting a to-be-processed rain image into the rain removal neural network model for image processing to obtain a rain-free clear image. According to the invention, rain stripes in a single rain-containing image can be effectively removed, and a clear image is obtained; and meanwhile, background information in the original image is fully reserved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a method for removing rain from a residual single image based on an attention mechanism, which can be used for sharpening processing of a single image. Background technique [0002] As one of the most important perception systems of human beings, vision is the main source of our knowledge. As pictures and videos spread more and more abundantly in the Internet age, images have become an important part of our lives. Therefore, computer vision is a very important subject in the context of today's era, and it is an important part of various fields in our lives, such as manufacturing, military industry, and various intelligent system fields. [0003] In a bad weather environment, the pictures or videos taken by people are often disturbed and blurred by rain, snow, and fog. The acquired picture or video information is damaged to varying degrees, and even the main...

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
CPCG06T5/003G06N3/084G06T2207/20081G06T2207/20084G06N3/048G06N3/045Y02A90/10
Inventor 吴炜汪萍
Owner XIDIAN UNIV
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