A Study of Predictable Reversible and Distortion-free Image Data Hiding Algorithms

• Main Authors: Chung-Min Yu, 游宗旻
• Other Authors: Chung-Ming Wang
• Format: Others
• Language: en_US
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/5m45b3

博士 === 國立中興大學 === 資訊科學與工程學系所 === 99 === The intention of data hiding algorithms is to convey hidden messages inside a cover medium in such a way that no one, apart from the sender and intended recipient, suspects the existence of a concealed message within a stego medium. In this thesis, we propose three data hiding algorithms. For reversible data hiding using the difference expansion approach there are three problems that are not yet solved: how to efficiently produce a stego image that satisfies the user’s desired pure capacity and image quality; how to automatically achieve the highest embedding efficiency ratio to balance the pure capacity and the image quality for a given cover image; and, given an image database, how to efficiently select the most appropriate cover image that has the highest embedding efficiency ratio to convey the secret messages without trying each of them. In the first algorithm, we introduce an efficient predictable scheme with the time complexity of O(1) to solve these three problems. By this method, we can foretell the embedding efficiency without real embedding. This provides an advantage when handling large data embedding systems such as an image database. Experimental results verify that our predictable scheme is precise where the average error rates are 2.63% for predicting the highest embedding efficiency, 1.52% for forecasting pure capacity, and finally, 0.3% for foreseeing image quality. We conclude that our prediction scheme is simple, precise, feasible, and efficient, and the average prediction time is around 15 milliseconds. We extend our first algorithm to support the reversibility when employing high dynamic range images such as our cover media. The high dynamic range image of interest is encoded by the red-green-blue-exponent encoding format. We present an E-value adjustment algorithm for secret message embedding taking advantage of the encoding format. The algorithm allows us to find a closer pixel value that greatly reduces the image distortion due to the hidden message. As a result, our algorithm produces a stego image with better image quality when the user demands the capacity of over 0.3 bits per pixel. The experimental results verify that our E-value adjustment algorithm has a better performance than that collected using a naïve difference expansion approach which indicates of using the difference expansion on the red-green-blue channels only regardless of the exponent channel encoding. . The third algorithm we present is a distortion-free data hiding method. Our scheme provides three significant benefits. First, it enables us to convey secret messages to produce a stego HDR image. To the best of our knowledge, our algorithm is the first approach in HDR literature that can provide capability of distortion-free data embedding. When we operate the tone mapping technique to reduce the high contrast to a displayable range, no distortion is encountered between the tone-mapped cover and the stego images. A quantitative measure verifies that histograms of the cover and stego HDR images are correlated with linear dependency. Second, our algorithm performs with adaptive message embedding where pixels conceal different amounts of secret messages based on their homogeneous representations. Quantitative analysis indicates that our algorithm offers an insignificantly small magnitude of the maximal pixel difference between the cover and stego HDR images. This feature and the histogram distribution of similarity between the cover and stego HDR images increase the difficulty of detecting whether any message is hidden in an HDR image. Third, our scheme is efficient. The time required for message embedding or extraction is in the range of several hundred milliseconds. Our approach belongs to a blind detection, and we believe our proposed scheme is suitable for applications such as image annotation or image steganography. In conclusion, we believe the three algorithms we propose extend the feasibility and applications for conveying secret messages through the data hiding approach.