Single image weak supervision defogging method based on priori knowledge and deep learning

A technology of prior knowledge and deep learning, applied in the field of single image dehazing, can solve the problems of not covering foggy scenes, poor effect, difficulty in training data for foggy/non-fog images, etc. Simple, enhance the effect of dehazing

Active Publication Date: 2021-02-26
ROCKET FORCE UNIV OF ENG
View PDF6 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The above method needs to use pictures of known depth to synthesize training data sets during model training, but these synthetic data cannot cover various foggy scenes in practical applications, so it is applied to some dense fog scenes or scenes without similar training data less effective
In addition, it is very difficult to collect pairs of foggy / nonfoggy image training data in many specific situations

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 weak supervision defogging method based on priori knowledge and deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] Such as figure 1 As shown, a single image weakly supervised dehazing method based on prior knowledge and deep learning of the present invention comprises the following steps:

[0026] Step 1. Establish a foggy image training set: Use prior knowledge to collect real foggy image samples {X i＝1...N} to carry out preliminary dehazing, and obtain the training sample set of fogged images {X i＝1...N ,Y i＝1...N}, where Y iRepresents a foggy image with X i The corresponding prior knowledge dehazes the image, and N represents the number of training samples;

[0027] Step 2: Build a weakly supervised dehazing model, the process is as follows:

[0028] Step 201, use convolution, batch normalization and activation function to form a convolution block, after continuous convolution coding, obtain the feature map f of the original input image size 1 / 16 1 / 16 ;

[0029] Step 202, for f 1 / 16 After pooling and upsampling at 1 / 2, 1 / 4, 1 / 8, and 1 / 16 scales, respectively, the features...

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 weak supervision defogging method based on priori knowledge and deep learning. The method comprises the steps of 1, establishing an atomized image training set;2, constructing a weak supervision defogging model; 3, defogging training of the foggy image pair; 4, calling a new foggy image pair, and stopping the training until the number of training steps or the loss value is not reduced any more or the loss value is less than 0.001 in the training process; and 5, defogging a single actual foggy image. A defogging result can be obtained by using a traditional priori knowledge method without training samples, although the result is not an optimal result, the result can be used as weak supervision information, so that the deep defogging model can better carry out weight learning, a large number of training images do not need to be synthesized, and defogging of a single image is realized.

Description

technical field [0001] The invention belongs to the technical field of single image defogging, and in particular relates to a weakly supervised single image defogging method based on prior knowledge and deep learning. Background technique [0002] Images collected under severe weather such as fog and haze will suffer from quality degradation due to atmospheric scattering, which will make the image color off-white, reduce the contrast, and make it difficult to identify object features. It can also lead to biased understanding of image content. Image defogging refers to the use of specific methods and means to reduce or even eliminate the adverse effects of suspended particles in the air on images. Single image defogging refers to dehazing to obtain a clear image under the condition of only one foggy image. At present, single image defogging methods are mainly divided into three categories: the first category is based on image enhancement methods, the second category is base...

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): G06T5/00G06K9/62
CPCG06T5/003G06T2207/10004G06T2207/20081G06T2207/20084G06F18/24147G06F18/241Y02A90/10
Inventor 苏延召崔智高姜柯
Owner ROCKET FORCE UNIV OF ENG
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