| Summary: | 碩士 === 國立臺中技術學院 === 資訊科技與應用研究所 === 95 === Recently, protecting the intellectual property rights of digitized information is a serious challenge. For the reason above, related information hiding technologies are becoming more and more important. In accordance with different requests, there are three different schemes proposed in this thesis.
The first scheme presents a new block-based authentication watermarking for verifying the integrity of binary images. The original protected image is partitioned into individual blocks. Each block obtains the hashing message by a hashing function. An exclusive-or operation is performed on the hashing message and watermark values and thus the authentication information is embedded into the protected image. If a binary image is tampered with by random modification or counterfeiting attack, the proposed technique can detect which locations have been altered.
In many data hiding techniques, the simple least-significant-bit (LSB) substitution is a general scheme used to embed secret message in the cover image. This practice may injure the quality of the host image which increases the probability that malicious users will notice the existence of something within the stego-image. As a result, the optimal LSB substitution method was proposed to improve the quality of the image, but the optimal LSB substitution solution is not easy to find. Therefore, the second scheme proposed an efficient algorithm as an attempt to solve the above problem. In the second proposed scheme, the optimal LSB substitution problem is regarded as a general assignment problem, and then the Hungarian algorithm is used to find the actual optimal LSB substitution solution. Also the proposed scheme does not need a great deal of memory space.
The third scheme proposed an effective reversible steganographic technique. The main concept is to utilize a similar property of all neighboring pixels. In the proposed scheme, the cover image is divided into non-overlapping groups by the neighboring pixel. Then each group is counted with an error value and then a complete error table can be derived. Then, the frequency of each error value is summed up and allows for the construction of the error histogram. Finally, the histogram shift scheme is used to hide data. The experimental results prove that by using the proposed scheme, the payload size and covered image quality are both obviously better than the original histogram shift scheme.
|
|---|
