Image Denoising Based on K-means Singular Value Decomposition

Ren, Jinchang; Lü, Hua; Zeng, Xiliang

doi:10.12928/telkomnika.v13i4.1897

Cited by 5 publications

(3 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To enhance the sparsity of the damaged image, in this section, the adaptive over-complete dictionary is employed to substitute the traditional sparse transform basis, that is, the image blocks will be represented by trained KSVD method [ 36 ], i.e.,

. It should be noted that the matrix A may not satisfy the RIP criterion, due to the over-complete dictionary D ; to address this issue, the mutual coherence technique [ 37 ] is applied to substitute the matrix A , i.e.,

where

and

are the i -th and j -th column in matrix A , respectively.…”

Section: Sparse Representation and Ksvd Algorithmmentioning

confidence: 99%

Microstructure Images Restoration of Metallic Materials Based upon KSVD and Smoothing Penalty Sparse Representation Approach

Liang

2018

Materials

View full text Add to dashboard Cite

Microstructure images of metallic materials play a significant role in industrial applications. To address image degradation problem of metallic materials, a novel image restoration technique based on K-means singular value decomposition (KSVD) and smoothing penalty sparse representation (SPSR) algorithm is proposed in this work, the microstructure images of aluminum alloy 7075 (AA7075) material are used as examples. To begin with, to reflect the detail structure characteristics of the damaged image, the KSVD dictionary is introduced to substitute the traditional sparse transform basis (TSTB) for sparse representation. Then, due to the image restoration, modeling belongs to a highly underdetermined equation, and traditional sparse reconstruction methods may cause instability and obvious artifacts in the reconstructed images, especially reconstructed image with many smooth regions and the noise level is strong, thus the SPSR (here, q = 0.5) algorithm is designed to reconstruct the damaged image. The results of simulation and two practical cases demonstrate that the proposed method has superior performance compared with some state-of-the-art methods in terms of restoration performance factors and visual quality. Meanwhile, the grain size parameters and grain boundaries of microstructure image are discussed before and after they are restored by proposed method.

show abstract

where

and

are the i -th and j -th column in matrix A , respectively.…”

Section: Sparse Representation and Ksvd Algorithmmentioning

confidence: 99%

Microstructure Images Restoration of Metallic Materials Based upon KSVD and Smoothing Penalty Sparse Representation Approach

Liang

2018

Materials

View full text Add to dashboard Cite

show abstract

“…If the image size is large, it will cause the gray level to overlap between foreground and background document. Meanwhile, Another algorithm that has been done is the K-Means Singular Value Decomposition (K-SVD) algorithm or by Ren, Lu, and Zeng (2015). The principle of this algorithm is to train a dictionary that represents the semantic structure of the image based on the library of the original image.…”

Section: Figure 1 Example Of Input Imagementioning

confidence: 99%

Filtering and Wavelet Transform Algorithm for Old Document Image Restoration

Samosir

2017

ComTech

View full text Add to dashboard Cite

The aim of this research was to develop image restoration system using filtering and wavelet transform algorithm. Data collection was through observation and system was developed using prototyping model. Result of this research is a computer based on system to restore image containing noise. Based on the research process, filtering and wavelet transform algorithm can used to restore old document image from interferences (noise).

show abstract

“…Compared with other research papers concerned in de-noising image [18][19], our contribution to analyze the data during transmission provides a good way to understand how the noise attack the data, also it tell us what is the most sensitive data which make the difference in the quality of the received image, this can help us to propose some techniques that can protect the important data by using channel coding or by sending additional data that can help us in restoration.…”

Section: The Application Of Unequal Error Protection In Levelsmentioning

confidence: 99%

Data analysis for image transmitted using Discrete Wavelet Transform and Vector Quantization compression

Khelifi¹,

Lakhdar²,

Elawady³

2016

TELKOMNIKA

View full text Add to dashboard Cite

In this paper we are going to study the effect of channel noise in image compressed with vector quantization and discrete wavelet transform. IntroductionIn recent years lot of research has been done to enhance the quality of received image after transmission in noisy channel, this problem can be solved using different techniques (source coding, channel coding , joint source channel coding or by optimize in restoration filters) as shown in articles [1][2][3].A good representation and compression of image can make the difference on the quality of received image, also to understand how the channel noise attack the transmitted data can be provide an efficient techniques with less consumption in resource (processing unit, time, data size, …) and bandwidth usage.In image compression, we know that, the combination between vector quantization and discrete wavelet transform provide a good representation of reconstructed image. Since the discrete wavelet transform analyze image and give us lot of data about 2D signal which make the difference in the quality of reconstructed image however In case of transmission the compression using VQ and DWT is so sensitive to the channel noise as shown in the articles [4][5].To solve this problem we are going to analyze data during transmission to discover what is the most important parts that can make the difference in the quality or received image by doing lot of tests in transmitted indices according to the decomposed levels provided by discrete wavelet transform.The paper is organized as follow: In the first part we will introduce the compression using vector quantization and discrete wavelet transform. The second part the effect of the channel noise in image compressed with vector quantization and discrete wavelet transform. In third part the effect of the channel noise in each level generated by discrete wavelet transform with different BER. In the fourth part we will divide each level to eight sublevels finally we try to propose some solutions according to the orbited results to that can enhance the quality of received image.

show abstract

Image Denoising Based on K-means Singular Value Decomposition

Abstract: Abstract

Cited by 5 publications

References 16 publications

Microstructure Images Restoration of Metallic Materials Based upon KSVD and Smoothing Penalty Sparse Representation Approach

Microstructure Images Restoration of Metallic Materials Based upon KSVD and Smoothing Penalty Sparse Representation Approach

Filtering and Wavelet Transform Algorithm for Old Document Image Restoration

Data analysis for image transmitted using Discrete Wavelet Transform and Vector Quantization compression

Contact Info

Product

Resources

About