De-noising Medical Images Using Machine Learning, Deep Learning Approaches: A survey

2020 ◽  
Vol 16 ◽  
Author(s):  
Ali Arshaghi ◽  
Mohsen Ashourian ◽  
Leila Ghabeli
Keyword(s):  
Wavelet Transform ◽  
Medical Images ◽  
Median Filter ◽  
Linear Method ◽  
Wiener Filter ◽  
Learning Approaches ◽  
Local Means ◽  
Non Local ◽  
Salt And Pepper ◽  
Better Than

Objective: Several de-noising methods for medical images have been applied such as Wavelet Transform, CNN, linear and Non-linear method. Methods: In this paper, a median filter algorithm will be modified and explain the image de-noising to wavelet transform and Non-local means (NLM), deep convolutional neural network (DnCNN) and Gaussian noise and Salt and pepper noise used in the medical skin image. Results: PSNR values of CNN methods is higher and better than to others filters (Adaptive Wiener filter, Median filter and Adaptive Median filter, Wiener filter). Conclusion: De-noising methods performance with indices SSIM, PSNR and MSE are tested and survey the result of simulation image de-noising.

scholarly journals Noise Reduction in Medical Images using an Unbiased Non-local Means Method

IJARCCE ◽  
2017 ◽  
Vol 6 (5) ◽  
pp. 702-706
Author(s):  
Sushma C ◽  
Kavitha G
Keyword(s):  
Noise Reduction ◽  
Medical Images ◽  
Local Means ◽  
Non Local

A MODIFIED ADAPTIVE MEDIAN FILTER METHOD AND ITS APPLICATIONS IN MEDICAL IMAGES

2010 ◽  
Vol 22 (06) ◽  
pp. 489-496 ◽  
Author(s):  
Mei-Sen Pan ◽  
Jing-Tian Tang ◽  
Xiao-Li Yang
Keyword(s):  
Medical Images ◽  
Median Filter ◽  
Window Size ◽  
Filter Method ◽  
Mr Images ◽  
Adaptive Median Filter ◽  
Local Window ◽  
Salt And Pepper ◽  
Image Detail ◽  
Gray Values

Since the medical image is usually corrupted by noise, the filter method is applied to remove the noise and improve the image quality. In this paper, a modified adaptive median filter method is proposed for filtering the medical images. When identifying noises, by selecting the maximum and the minimum gray values in the image as a criterion of judging the noise pixels, the probability that a nonnoise pixel is misjudged to be a noisy one is reduced, and the processing time for finding the maximum and minimum gray values in each local window is drastically decreased as well. When filtering the image, according to the noise granularity function (NGF) in a 3×3 window, the filtering window size is adaptively adjusted, then the median filter is used to eliminate the current noise-marked pixel in the median image (MI) generated by the adaptive median filter, and at the same time the noise mark is cancelled. The proposed method may both effectively remove the noises, and preserve image detail information well. The experimental results reveal that the proposed method is particularly effective in filtering the impulse noises, also called salt-and-pepper noises superimposed on images, including computed tomography (CT) and magnetic resonance (MR) images.

Natural image noise removal using non local means and hidden Markov models in stationary wavelet transform domain

2017 ◽  
Vol 77 (15) ◽  
pp. 20065-20086 ◽  
Author(s):  
Asem Khmag ◽  
Syed Abdul Rahman Al Haddad ◽  
Ridza Azri Ramlee ◽  
Noraziahtulhidayu Kamarudin ◽  
Fahad Layth Malallah
Keyword(s):  
Wavelet Transform ◽  
Hidden Markov Models ◽  
Markov Models ◽  
Hidden Markov ◽  
Noise Removal ◽  
Natural Image ◽  
Transform Domain ◽  
Stationary Wavelet Transform ◽  
Local Means ◽  
Non Local

scholarly journals DEVELOPMENT OF A SYSTEM FOR DIGITAL WATERMARKS EMBEDDING INTO IMAGES BASED ON DCT-LWT-SVD

2020 ◽  
pp. 23-31
Author(s):  
A.V. Akhmametieva ◽  
A.A. Baraniuk
Keyword(s):  
Wavelet Transform ◽  
Copyright Protection ◽  
Signal To Noise Ratio ◽  
Median Filter ◽  
Low Frequency ◽  
Digital Watermark ◽  
Signal To Noise ◽  
Actual Problem ◽  
Noise Ratio ◽  
Salt And Pepper

Copyright protection of digital content is a rather actual problem of humanity in the 21st century. Misuses of multimedia content is very common, and their number is growing with each passing day. One type of copyright protection is the embedding of digital watermark (DW) in the content. In this paper a new method of embedding digital watermark into image using discrete cosine transform, lifting wavelet transform (LWT) with maternal wavelet "Dobeshi-8" and singular coefficients decomposition is proposed. Embedding is performed into the first singular number of the low frequency wavelet transform region. As a digital watermark, we will use a grayscale image normalized to a range from zero to ten to provide a high peak signal-to-noise ratio (PSNR). The research analyzed the developed method: the method of embedding and detecting information was tested for its resistance to various types of attacks, namely: application of noise overlay (Gauss and pulse noise, "salt and pepper"), "unsharp" filter and median filter, and compression attack (with quality coefficients for a complete container from 60 to 100). As a result of the conducted testing, it was established that the method is quite resistant to all the attacks, except for the "unsharp" filtering (the resulting performance is not satisfactory). The method showed good results in peak signal-to-noise ratio - the average PSNR value is 50.5 dB, as well as high rates of similarity between the embedded DW and the extracted one - from 77% to 97.6% while saving the full container in a lossless format, and up to 53, 05 dB and 91.96% while saving the image in a lossless format (JPEG).

scholarly journals Reconstruction of Corrupted Image From Salt And Pepper Noise From Median Filter

2012 ◽  
pp. 4-6
Author(s):  
Vishal Gautam ◽  
Tarun Varma
Keyword(s):  
Median Filter ◽  
Median Filtering ◽  
Salt And Pepper Noise ◽  
Filtering Algorithm ◽  
Noise Density ◽  
Mean Filter ◽  
Salt And Pepper ◽  
Image Details ◽  
Lower Noise ◽  
Better Than

- Inthis paper,we propose an improved median filtering algorithm. Here, we introduced salt and pepper noise for the image corruption and reconstruct original image using different filters i.e. mean, median and improved median filter. The performance of improved median filter is good at lower noise density levels.The mean filter suppresses little noise and gets the worst results.The experimental resultsshow that our improved median filter is better than previousmedian filterfor lower noise density (upto 60%). It removes most of the noises effectively while preserving image details very well.

Sign in / Sign up

Export Citation Format

Share Document