Image noise level estimation method based on deep learning

An image noise and deep learning technology, applied in image enhancement, image analysis, image data processing and other directions, can solve problems such as inapplicability, and achieve the effect of accurate estimation, good estimation and strong robustness.

Inactive Publication Date: 2018-05-04
TIANJIN UNIV
View PDF6 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] Common noise estimation algorithms are designed for Gaussian

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 level estimation method based on deep learning
  • Image noise level estimation method based on deep learning
  • Image noise level estimation method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0037] The technical scheme adopted by the present invention is:

[0038] Step 1. Propose a signal dependent noise (SDN) model:

[0039] I=f(L I )

[0040] I N =f(L I +n s +n c )+n q

[0041] Among them, I represents an ideal noise-free image, I N Represents the noise image actually obtained by the CCD camera, f(·) represents the camera response function, Indicates that it depends on the light intensity L I Noise component, Represents the noise component that has nothing to do with the signal, n q Represents quantization noise. Since the intensity of quantization noise is smaller than other noises, this component can be ignored. N here s And n c The noise parameters are assumed to be

[0042] Step 2. Data preprocessing

[0043] Step 2.1: Obtain the BSD500 (Berkeley Image Segmentation) data set and download the noise-free image from the Internet as the original noise-free image set.

[0044] Step 2.2: Use the noise model proposed in step 1 to artificially add noise to the noise-free i...

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 belongs to the field of computer image processing and aims to provide a framework based on deep learning to effectively estimate an image noise level. For this purpose, a technical scheme is adopted. According to the technical scheme, an image noise level estimation method based on deep learning comprises the steps that 1, a signal dependent noise (SDN) model is proposed; 2, data preprocessing is performed; 3, a neural network structure is constructed; 4, the learning rate and momentum parameters of a network are set, a deep learning framework caffe is utilized to train the convolutional neural network till cost loss is reduced to a certain degree and training reaches the maximum number of iterations, a training model is generated, and a cost loss function is selected; and 5,an image with noise is input into the trained model, and a noise level function is output. The method is mainly applied to computer image processing.

Description

technical field [0001] The invention belongs to the field of computer image processing, and relates to the estimation of the noise level of a charge-coupled device and a convolutional neural network. Specifically, the convolutional neural network is used to extract image features, learn the signal-related noise distribution, and establish the mapping relationship between the end-to-end noise image and the corresponding noise level function. Background technique [0002] Charge-coupled Device (CCD), as the core device of a digital camera, will generate complex noise during the imaging process. This noise is not simply additive Gaussian White Noise (AWGN), but a signal-dependent noise (SDN) that depends heavily on signal strength. Knowing the noise level of CCD (Charge Coupled Device) can not only be used to evaluate the quality of the image sensor, but also can be used to adjust the parameters of many computer vision algorithms, so it is of great significance to accurately e...

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
IPC IPC(8): G06T5/00
CPCG06T5/002G06T2207/20081G06T2207/20084
Inventor 杨敬钰刘鑫宋晓林李坤
Owner TIANJIN 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