Unlock instant, AI-driven research and patent intelligence for your innovation.

Image denoising method and device based on neural network and image encoding and decoding method and device based on neural network image denoising

A neural network and image coding technology, which is applied in the field of image coding or decoding methods and devices including neural network image denoising, can solve the problems of storage resource consumption and heavy burden of denoising processing speed, to ensure image quality and perception, Effect of reducing scale and reducing noise

Active Publication Date: 2019-12-27
ZHEJIANG UNIV
View PDF8 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This brings a large burden on storage resource consumption and denoising processing speed

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 denoising method and device based on neural network and image encoding and decoding method and device based on neural network image denoising
  • Image denoising method and device based on neural network and image encoding and decoding method and device based on neural network image denoising
  • Image denoising method and device based on neural network and image encoding and decoding method and device based on neural network image denoising

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0063] This example provides a neural network-based video denoising method, including:

[0064] The pixel values ​​of the image area to be denoised and the reconstruction prediction residual value of the image area to be denoised are input to the convolutional neural network. The specific structure of the neural network is as figure 2 shown. Use the pixel value processing subnetwork to process the pixel values ​​of the image region to be denoised, and use the prediction residual value processing subnetwork to process the reconstruction prediction residual value of the image region to be denoised; use a mixed processing subnetwork Jointly process the output of the pixel value processing sub-network and the output of the prediction residual value processing sub-network, and each layer of the neural network of the mixed processing sub-network processes the input of the mixed processing sub-network or the output of the upper layer of the neural network of the sub-network. The o...

Embodiment 2

[0068] This example provides a neural network-based video denoising method, including:

[0069] The pixel values ​​of the image area to be denoised and the reconstruction prediction residual value of the image area to be denoised are input to the convolutional neural network. The specific structure of the neural network is as image 3 shown. Use the pixel value processing subnetwork to process the pixel values ​​of the image region to be denoised, and use the prediction residual value processing subnetwork to process the reconstruction prediction residual value of the image region to be denoised; use a mixed processing subnetwork Jointly process the output of the pixel value processing sub-network and the output of the prediction residual value processing sub-network, and each layer of the neural network of the mixed processing sub-network processes the input of the mixed processing sub-network or the output of the upper layer of the neural network of the sub-network. The ou...

Embodiment 3

[0073] This example provides a neural network-based video denoising method, including:

[0074] The pixel values ​​of the image area to be denoised and the reconstruction prediction residual value of the image area to be denoised are input to the convolutional neural network. The specific structure of the neural network is as Figure 4 shown. Use the pixel value processing subnetwork to process the pixel values ​​of the image region to be denoised, and use the prediction residual value processing subnetwork to process the reconstruction prediction residual value of the image region to be denoised; use a mixed processing subnetwork Jointly process the output of the pixel value processing sub-network and the output of the prediction residual value processing sub-network, and each layer of the neural network of the mixed processing sub-network processes the input of the mixed processing sub-network or the output of the upper layer of the neural network of the sub-network. The o...

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 provides an image denoising method and device based on a neural network and an image encoding or decoding method and device based on image denoising of the neural network. According to the image denoising method and device based on the neural network, a convolutional neural network model is adopted. The input of the neural network is the pixel value of a to-be-denoised image area andthe reconstruction prediction residual value of the to-be-denoised image area. According to the image encoding or decoding method and device based on neural network image denoising, a convolutional neural network model is adopted. The input of the neural network comprises a reconstruction prediction residual error and a reconstruction pixel value or a filtered reconstruction pixel value. By adding the reconstruction prediction residual input, the mapping relation between the noisy image pixel value and the noiseless image pixel value can be learned in a small network scale, so that the noisyimage pixel value is denoised, and meanwhile, the subjective quality of the image is ensured. The method can be applied to user-oriented image playing and image preprocessing in various computer vision fields.

Description

technical field [0001] The present invention relates to image processing technology, in particular to a neural network-based image denoising method and device, and an image encoding or decoding method and device including neural network image denoising. Background technique [0002] Image denoising is a classic problem in the field of image processing. The subjective quality of noisy images is poor, which affects people's perception. At the same time, noisy video will also affect the performance of other computer vision algorithms. Therefore, it is essential to filter the image. [0003] Existing filtering techniques include spatial domain filtering, transform domain filtering, dictionary learning, and deep learning. [0004] Spatial filtering directly processes the image pixel values ​​to be filtered. [0005] Transform domain filtering transforms the pixel values ​​of the image to be filtered to the transform domain, filters on the transform domain and then inversely t...

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): G06T5/00G06T9/00
CPCG06T9/002G06T5/70
Inventor 虞露李道文
Owner ZHEJIANG UNIV