| Summary: | 博士 === 國立中正大學 === 資訊工程研究所 === 100 === Image data hiding techniques are proposed for undetectable communication. The goal of the research in this field is to develop ways to embed more secret information into a cover image while minimizing distortion of the image so that the existence of the embedded secret information is imperceptible. In other words, the secret information can be transmitted safely if the stego image is indistinguishable from the cover image, allowing it to escape the best efforts of potential attackers.
However, even though the distortion is made as small as possible, some applications, such as those in the medical and military research fields, cannot tolerate any distortion at all. Although reversible data hiding schemes can restore the original image, the stego image still has a degree of distortion before the secret information is extracted. If the distortion appears in the crucial regions, it is possible to cause misjudgment while receiver takes the stego image as recovered one. To solve this problem, we are proposing an image data hiding strategy with restricted region protection. The users specify the crucial regions in advance, and modifications to these important regions are avoided. The proposed scheme outperforms traditional reversible data hiding schemes from the standpoints of flexible hiding capacity and the ability to control the quality of the image.
Due to the characteristics of a digital image, i.e., the ease with which it can be created and modified, the topic of how to ensure the integrity of an image has been of serious concern in recent years. Therefore, in the second scheme of this thesis, a fragile watermarking scheme based on chaotic maps and the Hamming code is proposed. This scheme can detect and locate any modification precisely, including the burst bits that occur during transmission in the protected image.
Although networking technologies are promoted as having incredible speed, the number of Internet users is certain to increase and, as a result, the usage of network bandwidth will also increase. In addition, the extensive utilization of high-resolution, digital images will also contribute to further heavy demands on network bandwidth. Several image compression schemes have been proposed to increase the efficiency of network bandwidth use. Among these schemes, vector quantization (VQ) is used most widely because it is efficient and effective. Therefore, in this thesis, the last two reversible data hiding schemes based on VQ images are proposed for solving both the Internet security problem and the bandwidth requirement problem concurrently. These two schemes allow the secret information to be embedded into the VQ-coded image, and they also allow the complete recovery of the original, VQ-coded image.
|
|---|
