Denoising system of medical x-ray image
An optical image and medical technology, applied in the medical field, can solve the problems of sensitivity, poor image oscillation method, no mixed noise, etc., to improve image quality, protect image edge detail information, and achieve the effect of denoising
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0026] During the collection, processing and transmission of medical X-ray images, due to the influence of various reasons, there will be more or less various noises and poor contrast. Among them, the type of noise is also relatively complicated. For the convenience of research, it can be divided into two parts: Gaussian noise and non-Gaussian noise. Medically, the requirements for image clarity are relatively high, so looking for a method that can eliminate two types of noise at the same time Yes X The key to optical image denoising. The non-Gaussian part here is mainly salt and pepper noise, and the Gaussian part is mainly granular Gaussian noise. When the decomposition module is processed, the non-Gaussian noise is mainly salt and pepper noise.
[0027] Since the non-Gaussian noise is randomly distributed h 1 (x)=(x 1 ,x 2 ,x 3 ,x 4 ,...,x n ), x i (i=1,2.3...n) are decomposed image pixels, and the arrangement of each noise point is irregular. When denoising, it wil...
Embodiment 2
[0031] When the decomposition module is processing, the decomposition module uses the improved median filter to filter out the non-Gaussian part of the noisy image. The specific operation method is:
[0032] a. Using a 3×3 odd-numbered template, within the filtering window, the target pixel H(i, j), and G(i, j) after median filtering;
[0033] b. MAX(H(i, j)) & MIN(H(i, j)) in the window are noise points, marked as 0, others are signal points, marked as 1. filter window such as figure 2 shown;
[0034] c. Observe λ i the value of (i=1, 2, 3, ..., 6);
[0035] where: λ 1 =2×H(i,j)-H(i-1,j)-H(i+1,j),λ 2 =2×H(i, j)-H(i, j-1)-H(i, j+1), λ 3 =2H(i,j)-H(i-1,j-1)-H(i+1,j+1),λ 4 =2H(i,j)-H(i+1,j-1)-H(i-1,j+1);
[0036] H(i-1, j-1)
H(i-1,j)
H(i-1, j+1)
H(i, j-1)
H(i,j)
H(i,j+1)
H(i+1, j-1)
H(i+1,j)
H(i+1, j+1)
[0037] select lambda 1 ~λ 4 average of Compared with the threshold T, when it is greater than or equal to the th...
Embodiment 3
[0042] When the reconstruction processing module is processed, the Gaussian noise is mainly granular Gaussian noise; because the traditional thresholding method is mainly hard thresholding and soft thresholding. The discontinuity of the hard threshold will cause a large mean square error and oscillation in the processed image. There is always a constant deviation between the wavelet coefficients estimated by soft threshold and the actual wavelet coefficients, which makes the processed image too smooth, especially the boundary of the image. The invention uses a new improved threshold value method, which makes up for the shortage of hard threshold value and soft threshold value, and better protects the edge information of the image.
[0043] The reconstruction processing module uses the improved wavelet threshold method to filter out Gaussian noise. The specific operation method is: use the time-frequency characteristics and multi-resolution of wavelet to decompose the noisy ima...
PUM
Abstract
Description
Claims
Application Information
- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
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