Enhanced LSB watermarking methods based on scanning patterns for authentication of medical images

摘要

This thesis investigated the trend in the current watermarking technology of medical images which features authentication, tamper localization, and recovery. Nowadays the information system is used widely in hospitals and clinical departments globally. It is able to help the clinical professionals doing diagnosis, smoothen the transmission process, and store the patients' data and medical images. Since the system handles thousands of crucial medical data, a security and authentication method is seen as imperative to protect the data. Three watermarking techniques were proposed toudenhance the fragile watermarking algorithm aiming at improving the numbering technique before embedding the watermark data. They are Authentication Watermarking for Grey-Scale Medical Images Using Spiral Manner Numberingud(SPIRAL-LSB), Authentication Watermarking for Grey-Scale Medical Images Using Hilbert Manner Numbering (HILBERT-LSB) and Authentication Watermarking for Color Medical Images Using Hilbert Manner Numbering (HILBERT-LSB-C). All proposed methods used unique styles of numbering and mapping pixels to guarantee a good performance of the authentication system. This can be done by spreading the numbered original data as far as possible from the original locations. The watermark generating and embedding phases used the same processes in all schemes. They produced the authentication data by using simple operations, which were parity bits check and comparison between average intensity data. The schemes embedded theudauthentication data in the least significant bit (LSB). The embedded authentication data in the same host image were utilized to localize any tamper using block-wise approach. The method is effective since it only required a secret key and a public, chaotic mixing algorithm to recover the attacked image.