Research On Medical Image Authentication Technology Based On Reversible Watermarking

Nowadays,the digital medical image has become more popular,it has been a mainstream.The computer science and technology has been widely applied to the Medical Imaging.Undoubtedly,it provides a very convenient technique for medical diagnosis,such as the electronic medical record,the electronic diagnostic image,or the telemedicine.It is very convenient to use the network to share the patient.Otherwise,patient case files can be stored in the database,this method eases the pressure of paper storage and facilitates the audit and inspection.However,any technology will bring convenience and disadvantage.Medical image and case files maybe tamper or delete maliciously.All of these will create many distrust factors in today's doctor-patient relationship and bring challenge in the authenticity and reliability of medical images.The reversible watermarking technology can solve the problem of high overhead in the encryption system,and it can solve the permanent distortion of the carrier image caused by the general digital watermarking,too.In this dissertation,we do the intensive study of the content authentication technology of medical images.We construct a reversible watermarking algorithm for content authentication and we construct an authentication scheme based on reversible watermark.The main contents of this paper are as follows:First,we propose a reversible watermarking algorithm based on wavelet transform.We use a rounding operation for the wavelet coefficient as a carrier signal,in order to embed the watermark information.The restoration secret key of the carry image will be stored into IPR.The key will be used for lossless recovery of the carrier image after extracting watermark.We propose an adjusted rounding operation for the wavelet coefficient in extracting process.Then,the watermark will be extracted lossless.Secondly,we propose a reversible watermarking algorithm based on wavelet transform for image content authentication.We extract features from the 2 X 2 blocks of the image to generate "authentication detection watermark" and "feature recovery watermark" in this algorithm.The "feature recovery watermarking"preserves block category information and some pixel details.The "authentication and detection watermark" retains the feature information used for authentication of the block.We do the wavelet transform for every block,thus,it can avoid the influence of tempering to the wavelet domain of the whole image.In the concerns about security,we use two different secret keys to encrypt these two watermarks with logistic maps.Therefore,the security of the watermarks are improved.We use a secret key string to do the pseudo random scrambling for the position of the blocks.It increase the randomness of the watermark embedding.In tampering detection,we use these two watermarks to detect and screen in different levels.It can improves the accuracy of detection and location.In the recovery of the tampering region,we use these two watermarks information to reconstruct the category and characteristic information of image block.Finally,we propose a medical image authentication scheme based on reversible watermarking and robust watermarking.The scheme takes advantage of the black background of medical images.It also performs partitioned operation on images,and different regions can achieve different functions.We use a watermarking to hiding the important information.We embed the watermark into the unimportant area by using the robust watermark technology of transform domain.We use the reversible watermark for content authentication that we have proposed,in order to protect the content of the important areas of the medical image.Then,it can be guaranteed to recover the image in the important area lossless after extracting the watermark.We can learn something in the experimental results.On the one hand,the reversible watermarking algorithm for image content authentication in this paper can detect and locate accurately for the general tampering,constant mean change or paste tampering.It can also restore the approximate original image information with high quality.In the case of no tampering attack,the algorithm proposed in this paper can achieve reversibility and security.On the other hand,the medical image authentication scheme we proposed can also be interconnected with the security database,thus,it can resist the noise attack and detect,locate and restore the tamper attack area.It has good operability,extensibility and compatibility.