Anti-forensics in Data Hiding of Multimedia Images

• Main Authors: Ya-Ting Chang, 張雅婷
• Other Authors: Shiuh-Jeng Wang
• Format: Others
• Language: en_US
• Published: 1030
• Online Access: http://ndltd.ncl.edu.tw/handle/984h86

碩士 === 中央警察大學 === 資訊管理研究所 === 102 === Data hiding, or steganography, of digital image is to embed special and important contents in specific cover image, so that users without access authority cannot discover the existence of embedded information, and achieve the goal of data protection while data transmission. According to various algorithm designing, there are three main research area of data hiding: spatial domain, compression domain, and frequency domain. The main idea of image processing in spatial domain is to handle the pixels directly by modifying pixel values. In the compression domain, the pixel values are generally encoded to another format and then compress or embed information. The technique of data hiding in frequency domain transforms pixel values into a set of coefficients first and then uses the coefficients to carry secret data. No matter which domain of data hiding, there are cover media, secret messages, and output stego media or codestream. Another novel research is steganalysis, which aims to analysis an unknown image, and identify stego images or detect the embedded secret messages further. In this thesis, new techniques of steganography and steganalysis are proposed. First, we propose a reversible data hiding method in spatial domain which takes advantages of image interpolation and histogram modification. Second, a data hiding method in compression domain is proposed using vector quantization and enhanced the neighboring correlation technique. Third, a steganalysis for vector quantization-based data hiding is presented. We design algorithm for detect a classical data hiding method which split codebook, and identify the stego image. Comparing with prior works, our proposed steganography methods enlarge the embedding capacity, and improve the image quality or increase compression ratio, so that conceal the existence of secret messages. Our proposed steganalysis method is able to identify vector quantization-compressed images, and detect the stego image using the split codebook vector quantization data hiding.