Research Of Reversible Watermarking Algorithm For Image Based On Prediction Error

The development of information technology makes people enjoy the convenience of digital information, but also mired in the increasingly serious security risks of digital products illegally obtaining and copying. Digital watermarking is a method of embedding some authentication information in the digital carrier to identify and protect the authenticity and legality of the carrier while not affect the value of the host. For a long time, the digital watermarking technique in the aspect of image is a hot research. Taking digital image as carrier, this paper mainly studied the reversible watermarking which is an important branch in the field of digital watermarking. Reversible watermarking is digital watermarking technique which can extract watermark information while completely restore the information of the original host image. This technique is widely used in the fields which have higher requirements of the original carrier image such as medical, laws and military.After checking a huge amount of information, This paper had a comprehensive analysis and research of reversible watermarking. The main contents of the research include the following aspects.(1) Proposed a reversible watermarking algorithm based on block difference expansion. Difference expansion algorithm is the ideological basis of the prediction error expansion algorithm. The algorithm extends the difference between two adjacent pixels, and then embeds the watermark into the least significant bit. The difference positioning map of Tian classic difference expansion algorithm is a little large which affect the embedding capacity of watermark. In order to improve this shortcoming, this paper divided the image into different parts, and respectively selected the appropriate pixels block of each part. Based on these measures, the size of the positioning map is effectively reduced. The result of experiment shows when embedded the same capacity of watermark the algorithm proposed in this paper has lower distortion than Tian proposed algorithm.(2) Proposed a prediction error expansion algorithm based on sorting and adaptive prediction. Due to the general prediction error expansion algorithm precise is not enough in prediction operator, this paper proposed a prediction operator which has adaptive ability to the context of the image texture features. This operator is able to more fully utilize the redundant pixel space and expand the embedding capacity; and at the same time sort the smoothness value of the image pixels, thus be able to select more suitable embedding position. All of these can reduce the possibility of objective distortion caused by embedding watermark. Experimental results show that, compared with other algorithms, this algorithm can increase the embedding capacity, while effectively reduce the distortion of the image which contains watermarks.(3) Proposed a reversible watermarking algorithm of color image based on prediction error. This reversible watermarking algorithm is proposed by analyzing the correlation between each color component of the color image and combining with the prediction error expansion algorithm with adaptive capacity which mentioned here in before. Experimental results show that this algorithm can ensure the quality of image, while successfully extract the watermark and recover the host image without distortion.