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

An Image Dehazing Method Based on Multi-Scale Densely Connected Networks

A connected network and multi-scale technology, which is applied in the field of image defogging based on multi-scale densely connected networks, can solve the problems of reduced performance of defogging, limited robustness, poor effect, etc., to improve contrast and improve defogging Effect, deep learning, simple and effective effect of defogging

Active Publication Date: 2022-04-05
福建帝视科技集团有限公司
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

When the traditional method is used for image defogging, if the generation of foggy images is inconsistent with the prior conditions or assumptions of the algorithm, it will lead to a decrease in its defogging performance
When using a deep learning-based dehazing algorithm, due to its robustness limited by the data set, it will not work well when processing some images

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
  • An Image Dehazing Method Based on Multi-Scale Densely Connected Networks
  • An Image Dehazing Method Based on Multi-Scale Densely Connected Networks
  • An Image Dehazing Method Based on Multi-Scale Densely Connected Networks

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0058] Such as Figure 1-8 As shown in one of them, the present invention discloses an image defogging method based on a multi-scale dense connection network, which is divided into the following steps:

[0059] Step 1, training data preparation phase.

[0060] Described step 1 specifically comprises the following steps:

[0061] Step 1.1, select the training data set. What the present invention uses is the match data of CVPR NTIRE2018 Outdoor Dehaze, and it contains fog image and non-fog image pair. Wherein, the foggy image is formed by a certain algorithm from the fogless image.

[0062] In step 1.2, the image database is preprocessed to form a paired set of haze sub-images and high-definition haze-free sub-images. Adaptive histogram equalization method based on contrast limitation [9] Preprocess the foggy image I(x) to obtain the contrast-adjusted foggy image I(x 1 ). From the foggy image I(x 1 ) in the screenshot sub-image I by d*d (d=256 in the present invention) ...

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 an image defogging method based on a multi-scale dense connection network, which reconstructs images with different degrees of fog into relatively clear images, and greatly improves the image quality and visual experience. For the first time, the adaptive histogram equalization method is proposed to improve the image contrast of the image preprocessing, which significantly improves the dehazing effect; the multi-scale densely connected convolutional neural network can describe the characteristics of fog at different scales, and effectively combine its characteristics. The most effective defogging effect is achieved; a formula based on the Retinex defogging problem is proposed, which makes the end-to-end deep learning defogging more concise and effective; the present invention is compared with other deep learning-based defogging algorithms. The invention not only greatly reduces the number of model parameters, but also can achieve an ideal defogging effect under the condition of very little training data.

Description

technical field [0001] The invention relates to the field of image enhancement, in particular to an image defogging method based on a multi-scale densely connected network. Background technique [0002] Fog is a weather phenomenon formed by the accumulation of tiny dust and water vapor particles in dry conditions. Turbid media such as fog, haze, and smoke will absorb and scatter atmospheric light, which will degrade images of outdoor scenes collected under this weather. Typically, degraded images lose contrast and color fidelity. [0003] The light intensity in the original direction will be gradually weakened by the light passing through some kind of scattering medium, and at the same time, due to the law of energy conservation, the weakened light intensity will be scattered to other directions. Also, the energy lost by scattering depends on its distance from the camera. Based on this physical phenomenon, people often use a physical model based on atmospheric scattering ...

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 Patents(China)
IPC IPC(8): G06T5/00G06T5/40G06N3/04G06N3/08
CPCG06N3/084G06T5/003G06T5/40G06T2207/10024G06T2207/10004G06T2207/20084G06T2207/20081G06N3/045
Inventor 刘文哲李根童同高钦泉
Owner 福建帝视科技集团有限公司
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