Reversible Information Hiding Based On Integer Wavelet Transform

With the rapid development of internet technology, it has become easier to make, copy, transmit and distribute multi-media digital works through internet. Among those technologies, Information Hiding Technology provides a very effective and reliable solution to the problem of multimedia information security and intellectual property protection. As this technology is under constant improvement, it turns to be more perfect.Concerning the other studies of information hiding algorithm, the original vector data cannot be recovered without any damage after the original vector image is embedded in the hidden information. This weakness is not allowed to appear in the fields of medicine and military, for those fields require high quality of original data. Reversible Information Hiding Technology comes into being for meeting the need of such requirement. In essence, this technology refers to a fragile watermarking scheme. Through this technology, the embedment of secret information in the multimedia works doesn't bring any damage to the original image. When the data of the image is doubted, secret information can be extracted from the multimedia work to identify the authenticity of this image. Besides, this technology is able to recover the original image without any damage.This thesis begins with an introduction to some background information of this technology, meaning of study, relevant concepts, classification, general model, the typical algorithms, performance evaluation, evaluation criteria, future research, research status and current problems. Further introduction goes to the basic theories of wavelet transform including the continuous wavelet, discrete wavelet, integer wavelet. Finally, much emphasis is put on how to improve the embedding capacity and the visual quality.The main contribution of this thesis lies in:(1) It has proposed a reversible information hiding algorithm which is based on predicting secondary expansion and adjustment of pixel value. The selection of prediction predictor diamond pixels ensures the highest accuracy of prediction. And the prediction error is subdivided into two categories: transformation and non-transformation. The generation of non-transformational distribution table relies on the transformability of the prediction error. Being the distribution table, the length of one-dimensional array is 256, which is used to record the corresponding non-transformable different amount of each predictive error value. When the amount equals to 0, the predictable error value is transformable. In addition, data can be extracted or predictable error value can be recovered. When the value is not equal to 0, the possibility of transformation is uncertain, for it may contain a convertible predictable error. In order to avoid the loss of embedding capacity, the second pixel value adjustment method is further introduced. This method is applied to figure out the predictable error. Since the introduction of the non-transformable error distribution table and the adjustment methods of the pixel values do not involve the embedment of binary location map, the embedding capacity is accordingly enlarged. It can be concluded from the experiment that the algorithm ensures good visual quality and higher embedding capacity.(2) By employing algorithm, a Reversible information hiding system is created on the basis of Visual Studio 2008. This system is only operated under the original carrier and the secret information in gray image. The result shows that such kind of system can be applied to embed, inspect, draw, authorize, remove secert information as well as make robustness testing and performance testing.