Steganography Detection System Based on Neural Network
碩士 === 國立中興大學 === 資訊科學研究所 === 92 === The purpose of this thesis is to advance a new steganography detection system. By adopting the features of back propagation learning neural network (BPN) and Principle Component Analysis (PCA), the proposed system provides the ability to indicate wheather an imag...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Online Access: | http://ndltd.ncl.edu.tw/handle/82780342567508561329 |
id |
ndltd-TW-092NCHU0394042 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-092NCHU03940422016-06-17T04:16:22Z http://ndltd.ncl.edu.tw/handle/82780342567508561329 Steganography Detection System Based on Neural Network 植基於類神經網路之影像隱藏資訊偵測系統 Yao-Te Hwang 黃耀德 碩士 國立中興大學 資訊科學研究所 92 The purpose of this thesis is to advance a new steganography detection system. By adopting the features of back propagation learning neural network (BPN) and Principle Component Analysis (PCA), the proposed system provides the ability to indicate wheather an image has been embedded other information or not. It works as follows:First, Discrete Wavelet Transform (DWT) decomposes a gray image into various frequency bands. Then the principal components of each band are fed into the neural network with PCA during batch process. We would like to emphasis that the use of PCA can simplify the amount of neurons for speeding the neural network. Finally, the general neural network with relations of principle components can determine the image is watermarked or not according to the training sample. In the simulation results, PCA not only reduces the training time but also reserves important characteristics in every gray level image. And the classification of these test images illustrates the combination of PCA and neural network provides well steganography detection ability. Gowboa Horng 洪國寶 學位論文 ; thesis 0 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立中興大學 === 資訊科學研究所 === 92 === The purpose of this thesis is to advance a new steganography detection system. By adopting the features of back propagation learning neural network (BPN) and Principle Component Analysis (PCA), the proposed system provides the ability to indicate wheather an image has been embedded other information or not. It works as follows:First, Discrete Wavelet Transform (DWT) decomposes a gray image into various frequency bands. Then the principal components of each band are fed into the neural network with PCA during batch process. We would like to emphasis that the use of PCA can simplify the amount of neurons for speeding the neural network. Finally, the general neural network with relations of principle components can determine the image is watermarked or not according to the training sample. In the simulation results, PCA not only reduces the training time but also reserves important characteristics in every gray level image. And the classification of these test images illustrates the combination of PCA and neural network provides well steganography detection ability.
|
author2 |
Gowboa Horng |
author_facet |
Gowboa Horng Yao-Te Hwang 黃耀德 |
author |
Yao-Te Hwang 黃耀德 |
spellingShingle |
Yao-Te Hwang 黃耀德 Steganography Detection System Based on Neural Network |
author_sort |
Yao-Te Hwang |
title |
Steganography Detection System Based on Neural Network |
title_short |
Steganography Detection System Based on Neural Network |
title_full |
Steganography Detection System Based on Neural Network |
title_fullStr |
Steganography Detection System Based on Neural Network |
title_full_unstemmed |
Steganography Detection System Based on Neural Network |
title_sort |
steganography detection system based on neural network |
url |
http://ndltd.ncl.edu.tw/handle/82780342567508561329 |
work_keys_str_mv |
AT yaotehwang steganographydetectionsystembasedonneuralnetwork AT huángyàodé steganographydetectionsystembasedonneuralnetwork AT yaotehwang zhíjīyúlèishénjīngwǎnglùzhīyǐngxiàngyǐncángzīxùnzhēncèxìtǒng AT huángyàodé zhíjīyúlèishénjīngwǎnglùzhīyǐngxiàngyǐncángzīxùnzhēncèxìtǒng |
_version_ |
1718307761409753088 |