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

Image noise estimation method based on deep convolutional neural network

A neural network and deep convolution technology, applied in the field of image noise estimation, can solve problems such as poor practicability, achieve good practicability, improve generalization ability, and improve flexibility

Active Publication Date: 2020-02-28
NORTHWESTERN POLYTECHNICAL UNIV
View PDF11 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the shortcomings of poor practicability of existing image noise estimation methods, the present invention provides an image noise estimation method based on deep convolutional neural network

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
  • Image noise estimation method based on deep convolutional neural network
  • Image noise estimation method based on deep convolutional neural network
  • Image noise estimation method based on deep convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0030] refer to figure 1 . The specific steps of the image noise estimation method based on deep convolution neural network of the present invention are as follows:

[0031] Deep Convolutional Neural Network Training with Noise Distribution and Noise Level Constraints:

[0032] (a) Build the training set:

[0033] Collect 500 pictures in any scene, require the images to be noise-free, expand the 500 pictures to 4000 pictures by rotating at any angle, 2-4 times reduction, etc., and further intercept 200×200 pixel texture density from each picture Moderate areas with complete texture structures, and finally get 4000 training picture sets with a size of 200×200 pixels;

[0034] All training pictures are divided into image blocks in an overlapping manner, each image block is 50×50 in size, and the centers of adjacent image blocks in the horizontal or vertical direction on the training picture are separated by 10 pixels. The image blocks obtained by all segmentation constitute ...

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 noise estimation method based on a deep convolutional neural network. The method is used for solving the technical problem that an existing image noise estimation method is poor in practicability. According to the technical scheme, a loss function is constructed by combining a noise error and a noise level error to train a deep convolutional neural network; the trained deep convolutional neural network obtains a noise map with accurate numerical distribution and accurate statistical mean square error from the noise pollution image; a deep convolutional neural network is adopted; convolutional kernel expansion values are firstly multiplied layer by layer and then symmetrically decreased layer by layer in an equal proportion mode, the extraction capacity of the deep convolutional neural network for random noise is improved, the number of convolutional kernels of each convolutional layer is large, a distribution model and the level of noise used in training are completely random, and the generalization capacity of the deep convolutional neural network for random noise is improved. According to the method, the noise image is extracted from the noise pollution image, the flexibility of modeling and analyzing the noise in subsequent operation is improved, and the practicability is good.

Description

technical field [0001] The present invention relates to an image noise estimation method, in particular to an image noise estimation method based on a deep convolutional neural network. Background technique [0002] Document 1 "Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising, IEEE Transactions on Image Processing, 2017, Vol26(7), p3142-3155" discloses a noise estimation based on deep convolutional neural network for image restoration method, this method learns a deep residual convolutional neural network, which can estimate the numerical distribution image of the real noise from the noise-contaminated image to a certain extent, but this method serves to improve the quality of image restoration, and is dedicated to improving the quality of images under Gaussian noise pollution. The restored PSNR value leads to a large deviation between the estimated noise level and the actual noise level, and it only has a good effect on Gaussian noise. [0003]...

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): G06T5/00G06N3/08G06N3/04
CPCG06N3/08G06T2207/20081G06T2207/20084G06N3/045G06T5/70
Inventor 朱宇葛鑫姜伟孙瑾秋张艳宁
Owner NORTHWESTERN POLYTECHNICAL 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