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

A Quantization Method for Image Compression System Based on Dictionary Learning

A dictionary learning, image compression technology, applied in image communication, image coding, image data processing and other directions, can solve problems such as poor quantization performance

Active Publication Date: 2018-04-10
TSINGHUA UNIV +1
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the quantization performance of this method is poor when the data have different weights, especially when the weights of data-dense regions are small

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
  • A Quantization Method for Image Compression System Based on Dictionary Learning
  • A Quantization Method for Image Compression System Based on Dictionary Learning
  • A Quantization Method for Image Compression System Based on Dictionary Learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0092] Specific requirements:

[0093] use Figure 4 As shown, 10 natural images with a size of 512×512 are used to train the learning dictionary, and the learning dictionary is used to Figure 5 The shown natural image of size 512 × 512 was compressed using a "dictionary learning-based image compression system", requiring its reconstructed image PSNR to be at least 50dB.

[0094] Offline part:

[0095] Using the K-SVD dictionary learning algorithm to Figure 4 The 10 natural image data with a size of 512×512 are shown as training samples, and a learning dictionary D with a size of 64×512 is trained.

[0096] Online section:

[0097] On the encoding side, follow the steps below to Figure 5 The shown natural image data to be compressed is converted into a bit stream suitable for channel transmission:

[0098] Step (1), initialization:

[0099] Stored in the learning dictionary D,

[0100] Set the reconstructed image minimum The given value is 50dB,

[0101] The init...

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

A quantization method of "an image compression system based on dictionary learning" belongs to the field of image compression technology in multimedia communication, and is characterized in that the coefficient matrix is ​​removed from zero coefficients, the non-zero coefficient values ​​are sorted, and the estimated truncation coefficient percentage is used to truncate The non-zero coefficient sequence and the non-zero coefficient sequence retained by normalization are divided into equal sub-intervals by the uniform quantization method, and the K-means clustering and quantization is performed independently in each sub-interval, and the K-means clustering In the iterative process of class quantization, the mean value of all elements in each category is used as the new cluster center of the category. After the iteration termination condition is satisfied, all non-zero coefficients in each category are quantized to the corresponding cluster center values. Calculate the PSNR of the reconstructed image, and compare it with the set minimum PSNR value of the reconstructed image, adjust the truncation coefficient percentage, and repeat the above operation until the PSNR calculation value of the reconstructed image is not lower than the minimum PSNR specified value of the reconstructed image. Compared with uniform quantization or K-means clustering quantization alone, the present invention has the best quantization performance.

Description

technical field [0001] The invention belongs to the technical field of image compression in multimedia communication. Background technique [0002] Image compression has always been one of the classic problems in the field of image processing, aiming at removing redundancy and correlation in images to achieve efficient transmission or storage of image data. In order to meet the growing application requirements and the rapid development of multimedia communication technology, multimedia files such as videos and images continue to break through in the direction of higher precision and higher resolution. posed serious challenges. In order to effectively solve this problem, image compression has always been a research hotspot in the field of image processing. [0003] Image compression methods can be divided into two categories: lossless compression and lossy compression. The former requires the decoder to restore the original image without distortion, while the latter allows...

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): H04N19/124H04N19/13H04N19/94H04N19/91G06K9/00G06K9/62G06T9/00
CPCH04N19/124H04N19/13G06T9/008G06V40/16G06F18/2136G06F18/28G06F18/23
Inventor 陶晓明王隽徐迈刘喜佳葛宁陆建华
Owner TSINGHUA 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