Study on Novel Adaptive Data Hiding Schemes

• Main Authors: WENG, YI-CHUN, 翁奕鈞
• Other Authors: KUO, WEN-CHUNG
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/2dp7aa

碩士 === 國立雲林科技大學 === 資訊工程系 === 105 === With the rapid development of information technology, Internet has become the main communication channel between people. However, with the popularity of social networking sites and communication software, network security issues are gradually expanding. Protecting personal and account security becomes a very important issue because hackers may steal and alter personal information through various means. Data hiding is embedding the private message into the multimedia file to achieve secret communications between the sender and the receiver. There are three criteria in data hiding: capacity, image quality and security. Each paper researches in view of these criteria. In this thesis, we will present a new idea of adaptability, which makes a comparison and prejudgment before hiding secret messages to achieve better embedding capacity and image quality. Therefore, two data hiding schemes are proposed in this thesis. First, we use to the irregular selection method instead of the rule arrangement method in GEMD scheme. In other words, we expand the pixel group form n pixels to n or (n+1) pixels for prejudgment conditions, and then choose the condition of fewer modified values of pixels to embed. The stego-image quality of our proposed scheme is better than GEMD. Besides, it is more secure than fixed pixel group because the choice of n or (n+1) condition is random. The other, we propose a reversible data scheme to improve the typical reversible data hiding with difference prediction. After selecting the neighboring pixel values for the second prediction, we can adjust the larger difference between the neighboring values by using the principle of proportionality. According to our analysis, the predicted value is closer to the original pixel value and makes more embedding chances to increase capacity. Finally, we use the RS detection to judge our proposed scheme, in order to prove our scheme is not only to maintain the original security but also to improve the stego-image quality. Another our proposed reversible data hiding scheme can establish the difference prediction histogram to select the best proportion of the allocation and compare the number of peaks with traditional predictions. According to our simulation, the major contribution of this scheme is to enhance the rise of embedding capacity and maintain good stego-image quality.