Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Combined medical science image fusion denoising method based on discrimination dictionary learning

A medical image and dictionary learning technology, which is applied in the field of joint medical image fusion and denoising, can solve the problems of joint medical image denoising and fusion difficulties, achieve clear edge details, improve fusion effect, and avoid the effect of oscillation response propagation

Active Publication Date: 2018-01-09
KUNMING UNIV OF SCI & TECH
View PDF3 Cites 20 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by the present invention is to provide a method for fusion and denoising of joint medical images based on discriminative dictionary learning, so as to solve the problem of difficulty in denoising and fusion of joint medical images in the prior art

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
  • Combined medical science image fusion denoising method based on discrimination dictionary learning
  • Combined medical science image fusion denoising method based on discrimination dictionary learning
  • Combined medical science image fusion denoising method based on discrimination dictionary learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0033] Embodiment 1: as figure 1 As shown, the present invention proposes a method for joint medical image fusion and denoising based on discriminant dictionary learning. Firstly, the image used for training is collected, and then the collected image is trained to obtain a sparse dictionary and a low-rank dictionary, and then the sparse Image decomposition of dictionaries and low-rank dictionaries to obtain sparse elements and low-rank elements, add sparse constraints to sparse elements, and add kernel norm and weighted kernel norm constraints to low-rank elements to obtain low-rank coefficients and sparse coefficients respectively. The rank dictionary and the low-rank coefficients are iteratively updated to obtain the updated low-rank components, and the sparse dictionary and the sparse coefficients are iteratively updated to obtain the sparse components. Finally, the low-rank components and the sparse components are fused to obtain the final fused image.

[0034] Step1, firs...

Embodiment 2

[0082] Embodiment 2: 8 images are collected in the process of training the dictionary. The initial low-rank dictionary and sparse dictionary are obtained by passing the training sample image through K-SVD. According to the proposed dictionary learning algorithm iteratively updated to obtain the required sparse dictionary D s and a low-rank dictionary D l ; In the image fusion method when the input source image has no noise, the figure 2 The source image of the source image is separated by the existing RPCA algorithm to obtain the low-rank component and sparse component of the source image; the parameters involved in the fusion algorithm η1, η2, η3, η4, η5 are set to 1, 1, 1, 0, 1, The number of iterations in dictionary learning is 5. Through the calculation of the image fusion algorithm, the fusion result obtained in the implementation of the matlab software is Figure 4 The last image in Table 1 shows the evaluation index of noise-free source image fusion and traditional...

Embodiment 3

[0086] Embodiment 3: The method of dictionary training is the same as example 1. In the image fusion method, when the input source image is destroyed by Gaussian white noise, the image 3 The source image, η1, η2, η3, η4, η5 are set to 1, 1, 1, 1.3, 1 respectively. The number of iterations is 10. Other operations are the same as Example 1. The noise source image used is destroyed by Gaussian white noise with a noise level of 10 or 20, and the obtained fusion result is Figure 5 The last image in . Table 2 shows the evaluation indicators of Gaussian white noise source image denoising fusion and traditional methods, and the maximum value of each set of evaluation data is enlarged and bolded.

[0087]

[0088] Table 2: The average quantitative evaluation of 10 groups of medical image noise by different methods

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 relates to a combined medical science image fusion denoising method based on discrimination dictionary learning, which belongs to the technical field of digital image processing. Firstly, medical science images are collected. Secondly, the collected medical science images undergo training, obtaining a sparse dictionary and a low-rank dictionary. Thirdly, image decomposition is conducted on the sparse dictionary and the low-rank dictionary, obtaining sparse element and low-rank element. The sparse element is added with sparse constraint and weighted kernel norm constraint, and thelow-rank element is added with kernel norm, obtaining low-rank coefficient and sparse coefficient. Iteration is conducted on the low-rank dictionary and the low-rank coefficient, obtaining the updated low-rank component. Iteration is conducted on the sparse dictionary and the sparse coefficient, obtaining sparse component. Finally, the low-rank component and the sparse component are fused, obtaining the final fusion image. The invention is advantageous in that when source images have noise are input, the fusion effect can still be good, and thereby effect is obviously raised.

Description

technical field [0001] The invention relates to a method for combining medical image fusion and denoising based on discriminant dictionary learning, which belongs to the technical field of digital image processing. Background technique [0002] The emerging biomedical imaging technologies provide new tools for the imaging description of biological tissues, and play an important role in clinical diagnosis, surgery, radiotherapy, and efficacy evaluation. In order to synthesize the complementary information of multimodal medical images and provide a more sufficient and effective basis for clinical medicine, medical image fusion emerged as one of the post-processing techniques of medical images in the 1990s. Medical image fusion belongs to the interdisciplinary research field of medical imaging and image informatics, and its appearance has opened a new page for the development of molecular imaging. [0003] There are different classification methods for medical image fusion, wh...

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): G06T5/00G06T5/50G06K9/46G06K9/62
Inventor 李华锋和晓歌
Owner KUNMING UNIV OF SCI & TECH
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
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