Research On Reversible Digital Watermarking Algorithm For Medical Images

With the high popularity of the Internet,it is more convenient to obtain and transmit information,privacy protection in the network environment has been given more and more attention.The reversible digital watermarking method can recover the carrier without distortion after extracting secret information.Therefore,it has been used in military,medical,judicial and other fields which require high accuracy for carrier data.In the medical field,reversible watermarking technology is used to protect patients’ privacy by embedding electronic medical record and some related information into a medical image.For application scenarios like this,special performance of reversible digital watermarking methods is required,such as large embedding capacity and high-quality images.However,it is usually difficult to achieve both of them at the same time.In this paper,we propose two reversible digital watermarking algorithms.Both of them have large embedding capacity and high image quality.The main works are listed as follows:(1)A reversible digital watermarking algorithm based on integer wavelet high-frequency subband coefficients and prediction error expansion is proposed.This novel algorithm has two advantages.Firstly,the directed difference prediction is designed for integer wavelet detail subband coefficients.Secondly,the reduced difference expansion is improved,not only to increase the number of available prediction errors,but also to reduce the difference of subband coefficient caused by watermark embedding.Comparative experiments show that,with the same image quality,the embedding rate of the proposed algorithm is on average 0.16bpp higher than other algorithms.(2)A reversible digital watermarking algorithm based on image threshold segmentation and contrast enhancement is proposed.There are three advantages in the algorithm.Firstly,the Ostu threshold segmentation is applied on the gradient amplitude of the image to obtain the pixel threshold.After that,a two-direction scan is used to determine the to-be-embedded area.Secondly,in the part of histogram equalization,the histogram bins with pixel value between peak values and zero values are shifted.Therefore,overflow or underflow is avoided.Thirdly,in order to extract the watermarking and restore the carrier image without distortion,a little additional information is embedded with the watermarking.With the increasement of embedding loops,the length of additional information increases.When the number of peak value is less than the length of additional information,the end of embedding loop comes.Experimental results show that the algorithm can not only achieve larger embedding capacity but also obtain stego images with higher quality.