Research On High-fidelity Reversible Watermarking Based On Local Texture Features And Its Application In Medicine

With the rapid development of digital medical imaging technology,information exchanging between different medical institutions has be more convenient.At the same time,how to protect the private information of patient contained in digital medical images is attracting more and more attention.Medical information sensitivity also makes any changes to the original image are not allowed,and further increases the difficulty of medical image privacy information protection.As an important branch of digital watermarking technology,reversible watermarking technique can hide the private information of the patient in the medical image,and recover the original carrier image while extracting the watermark,has a wide application in the information sensitive areas such as medical image.Traditional reversible watermarking algorithm mainly focus on the smooth region of the image to achieve a better embedding performance but many medical images have rich texture regions,so it is necessary to make use of the local texture effectively.In this paper,we focus on improving the embedding performance of existing reversible watermarking method through the effective use of local texture information in medical images.In this paper,two methods of reversible watermarking based on local texture feature of medical image are proposed based on edge detection principle and payload partition strategy,based on the full study and analysis of the reversible watermarking method for prediction error extension technique.The main work is as follows:By analyzing the correlation principle of image edge detection,this paper proposes a prediction method of pixel value by classifying the local region texture information of medical images and then using the corresponding prediction model to predict the pixel value.A new reversible watermarking method is proposed by using the proposed prediction method combined with prediction error extension technique.By comparing with the existing reversible watermarking method,it is verified that the proposed method can achieve better embedding performance in prediction accuracy,embedding capacity and embedded distortion control.By studying the correlation between different color channels in medical images,a improved sequence sequencing strategy utlizing the similarity of texture information between the channels is proposed.On this basis,a novel payload partition method based on greedy strategy instead of the global search to construct the payload combination is proposed to improve the time efficiency of the exhaustive search method.The proposed method has achieved a embedding performance approximated to the optimal payload partition scheme,while greatly improved its time efficiency.