ﺑﺮﺭسی ﺟﺎﻣﻊ کاﺭﺍیی ﺭﻭﺵ ﻧﻬﺎﻥ کاﻭی مبتنی ﺑﺮ یاﺩگیری ﻋﻤﯿﻖ ﺩﺭ کشف ﺭﻭﺵ ﻫﺎی ﺣﻮﺯﻩ مکاﻥ

ﻧﻬﺎﻥ نگاﺭی، ﻫﻨﺮ مکاﺗﺒﺎﺕ پنهانی ﺍﺳﺖ که ﺩﺭ ﺁﻥ یک پیام ﺑﻪ ﺻﻮﺭﺕ ﻣﺨﻔﯿﺎﻧﻪ ﻣﻨﺘﻘﻞ میﺷﻮﺩ ﻭ ﻧﻬﺎﻥ کاﻭی، ﻫﻨﺮ کشف ﺣﻀﻮﺭ ﺍﻃﻼﻋﺎﺕ پنهاﻥ ﺍﺳﺖ. شبکهﻫﺎی عصبی پیچشی ﺑﺮﺧﻼﻑ ﺭﻭﺵﻫﺎی ﻧﻬﺎﻥکاﻭی سنتی، ﺑﺎ ﺍﺳﺘﺨﺮﺍﺝ ﺧﻮﺩکاﺭ ﻭیژگیﻫﺎ، ﻭﺟﻮﺩ ﺩﺍﺩﻩ ﺭﺍ ﺗﺸﺨﯿﺺ میﺩﻫﻨﺪ. ﺩﺭ ﻣﻘﺎﻻﺕ ﻣﺨﺘﻠﻒ، عملکرﺩ ﻣﺪﻝﻫﺎی ﻣﻮﺟﻮﺩ ﺑﺮ ﺭﻭی ﺗﻌﺪﺍﺩ ﻣﺤﺪﻭﺩی ﺍﺯ ﺭﻭﺵﻫﺎی ﻧﻬﺎﻥنگاﺭی ﺣﻮﺯﻩ مکاﻥ گزﺍﺭﺵ ﺷﺪﻩ ﺍﺳﺖ. ﻫﺪﻑ ﺍصلی ﺍین ﻣﻘﺎﻟﻪ، ﺍﺭﺍﺋﻪ یک شبکه عصبی پیچشی ﻭ ﺑﺮﺭسی ﺟﺎﻣﻊ عملکرﺩ ﺁﻥ ﺩﺭ کشف ﺭﻭﺵﻫﺎی ﺣﻮﺯﻩ مکاﻥ ﻣﺨﺘﻠﻒ ﺍﺳﺖ. ﻣﺪﻝ پیشنهاﺩی ﺍﺯ ﺳﻪ ﻗﺴﻤﺖ پیش پرﺩﺍﺯﺵ، ﻣﺎژﻭﻝ پیچشی ﻭ ﻃﺒﻘﻪﺑﻨﺪ تشکیل ﺷﺪﻩ ﺍﺳﺖ. ﺩﺭ ﻻیﻪ ﻣﺎژﻭﻝ پیچشیﻻیه ﺗﻤﺎﻡ ﻣﺘﺼﻞ تشکیلﺷﺪﻩ ﺍﺳﺖ. ﺍﺯ ﺭﻭﺵﻫﺎی ﺟﺎﺳﺎﺯی ﺩﺭ ﺑﯿﺖ کم ﺍﺭﺯﺵ، 3 ﺑﻠﻮک ﻭ ﻃﺒﻘﻪﺑﻨﺪ ﺍﺯ 5 ﺍﺯ ﺟﺎﺳﺎﺯی ﺩﺭ ﻣﻘﺪﺍﺭ ﺍﺧﺘﻼﻑ پیکسلﻫﺎ ﻭ ﺟﺎﺳﺎﺯی مبتنی ﺑﺮ ﺍیﺪﻩ تطبیقی ﺑﺮﺍی ﺗﺴﺖ ﺍﺳﺘﻔﺎﺩﻩ ﺷﺪﻩ ﺍﺳﺖ. ﺭﻭﺵ ﻭﺟﻮﺩ ﺩﺍﺩﻩﻫﺎی ﺑﺎ ﻃﻮﻝﻫﺎی حتی ﺑﺴﯿﺎﺭ کم ﺩﺭ ﺭﻭﺵﻫﺎی ﺩﻭ گرﻭﻩ 97٪ پیشنهاﺩی میﺗﻮﺍﻧﺪ ﺑﺎ ﺩﻗﺖ ﺑﺎﻻﺗﺮ ﺍﺯ ﺍﻭﻝ ﺭﺍ ﺷﻨﺎﺳﺎیی کند. عملکرﺩ ﺭﻭﺵ پیشنهاﺩی ﺩﺭ کشف ﺩﺭﺻﺪ ﺟﺎﺳﺎﺯیﻫﺎی ﺑﺴﯿﺎﺭ کم ﺭﻭﺵ تطبیقی ﺑﺎ ﺩﻗﺖ ﺑﺴﯿﺎﺭ ﻣﻨﺎﺳﺐ ﺍﺳﺖ ﻭ ﺍیﻦ ﻭیژگی ﻧﻘﻄﻪ ﺗﻤﺎیﺰ ﻣﺪﻝ پیشنهاﺩی ﻧﺴﺒﺖ ﺑﻪ ﺭﻭﺵﻫﺎی سنتی ﺍﺳﺖ. ﺯیرﺍ 70٪ ﺑﺎﻻی ﻣﻮﻓﻘﯿﺖ ﺭﻭﺵﻫﺎی ﺍﺳﺘﺨﺮﺍﺝ ﻭیژگی ﺩستی ﺑﻪ ﺩﻟﯿﻞ کم ﺑﻮﺩﻥ ﺗﻐﯿﯿﺮﺍﺕ ﻭیژگیﻫﺎی ﺁﻣﺎﺭی ﺩﺭ ﺳﻄﻮﺡ ﺟﺎﺳﺎﺯی پایین، ﺑﺴﯿﺎﺭ کمتر ﺍﺳﺖ.

a comprehensive evaluation of deep learning based steganalysis performance in detecting spatial methods

steganalysis is the art of detecting the existence of hidden data. recent research has revealed that convolutional neural networks (cnns) can detect data through automatic feature extraction. several studies investigated the performance of existing models using a limited number of spatial steganography methods. this study aims to propose a cnn and comprehensively investigate its efficiency in detecting different spatial methods. the proposed model comprises three modules: preprocessing, convolutional (five blocks), and classifier (three fully connected layers). the test results for the least-significant-bit (lsb) and pixel-value differencing (pvd) based methods indicate that the proposed method can detect data of even concise length with highaccuracy and a low error. the proposed method also detects complexity-based lsb-m (cbl) as an adaptive approach. lower embedding rates make this success even more impressive. manual feature extraction has much lower success rates due to low variations of statistical features at low embedding rates than the proposed model.