Research On Medical Image And Medical Volume Data Watermarking Algorithm Based On Zernike

With the advent of the era of big data,smart healthcare has become an information business that can help doctors effectively share information with other doctors.It also allows patients to upload their health data to the cloud,and supports sharing these data with doctors in hospitals for medical treatment,which greatly facilitates doctors’ diagnosis of unfamiliar patients.However,although smart medicine has brought high efficiency and convenience,there is also a problem of personal important information leakage.Once leaked,it will not only bring huge economic and credibility losses to hospitals and other relevant institutions,but also cause potential harm to patients,and even endanger the lives of patients,which will seriously damage the healthy development of the medical industry.Therefore,ensuring that patient privacy is not leaked and avoiding unnecessary medical information accidents has become an urgent problem to be solved.Digital watermarking technology is now widely used in the field of privacy protection of medical images,and this technology can protect the private data in the images.It can embed the patient’s personal privacy information into the corresponding medical image,and has the characteristics of not changing the original medical image and invisible to the human eye.In this thesis,based on Zernike moment sub-pixel edge detection and discrete cosine transform(DCT)feature extraction algorithm,zerowatermarking technology,cryptography perceptual hashing and other theories,a robust zero-watermarking algorithm for two-dimensional medical images and three-dimensional medical images(also known as medical volume data)is proposed to protect patient privacy information security.The main research work is as follows :(1)A robust zero-watermarking algorithm for medical images based on ZernikeDCT transform is studied.Firstly,the 32-bit frequency domain coefficient feature vector obtained by Zernike-DCT transform is used to process the image.The feasibility of this feature extraction method is proved by comparing the correlation coefficients of feature vectors of different medical images and similar medical images.The digital watermark is encrypted by Logistic chaotic mapping,and the encrypted watermark is embedded and extracted by zero watermarking technology.Finally,the robustness of the watermarked medical image under different kinds of attack experiments is studied.The NC values are1,0.79 and 0.92 respectively when attacked by JPEG compression quality of 10 %,clockwise rotation of 20 ° and magnification of 2 times.The results show that the algorithm has strong robustness against conventional attacks and geometric attacks.(2)A robust zero-watermarking algorithm for 3D medical images based on ZernikeDCT transform is studied.Firstly,the medical volume data is sliced,and each slice is subjected to Zernike moment edge detection and 3D-DCT transform perceptual hashing to obtain the medical volume data feature vector.The Logistic chaotic map is used to scramble the digital watermark.The feature vector and the encrypted watermark are embedded and extracted by the key obtained by the cryptographic Hash function.When subjected to 5 % noise intensity,15 % horizontal left shift and 50 % Z-axis shear attack,the NC values were 0.93,1 and 0.94,respectively.The experimental results show that the method has good anti-attack interference ability.Even under high-intensity conventional and geometric attacks,the method can still extract watermark information completely and clearly.