A new multi-frames watermarking technique for tamper detection and recovery

• Main Author: Khor, Hui Liang
• Format: Thesis
• Language: English
• Published: 2017
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://umpir.ump.edu.my/id/eprint/19535/1/A%20new%20multi-frames%20watermarking%20technique%20for%20tamper%20detection%20and%20recovery..pdf; http://umpir.ump.edu.my/id/eprint/19535/

Nowadays, medical image modalities have been digitalized and standardized into DICOM format which exhibit multi-frames characteristics; hence, multi-frames medical images have been widely used in clinical practice but it has been found that most of watermarking researches were conducted on single-frame medical image because their focus was on the intrinsic algorithm of watermarking scheme. Multi-frames watermarking could be processed sequentially but it is time-consuming. Watermarking process is the background work before diagnosis process starts; therefore, the processing time should be shortened in order to minimize patient’s waiting and treatment time in a hospital environment. The objectives of the research are to resolve the time constraint of sequential multi-frames watermarking process. The sequential multi-frames watermarking process could be speed up by two methods, first method was designed a speedy watermarking scheme algorithm (ROI-DR) in a single frame of medical image, and the second method was developed a new multi-frames watermarking scheme by adding parallelism component into sequential multi-frames watermarking process. ROI-DR speed up the watermarking processing time by eliminating the tamper localization features and replacing the whole ROI image immediately with the extracted ROI image at the first encounter of tampered pixel. The experiment results show that ROI-DR has similar PSNR value (~ 48 dB) and shorter processing time as compared to TALLOR and TALLOR-RS watermarking scheme. ROI-DR has an average speed up factor of 22.55 and 26.65 in watermarking embedding process; 21.89 and 42.79 in watermarking authentication process relative to TALLOR and TALLOR-RS respectively. The proposed multi-frames watermarking scheme was developed by utilizing multicores technology where medical images have been divided among cores and perform watermarking process in parallel. Two different watermarking schemes (TALLOR and the ROI-DR) have been incorporated into the proposed multi-frames watermarking scheme. Overall speed up factor for ROI-DR and TALLOR range from 0.91 ~ 2.16 and 5.21 ~ 6.60 respectively. The higher speed up factor in TALLOR indicated that the performance of the proposed multi-frames watermarking scheme is affected by the algorithm complexity of the adopted watermarking scheme and the number of frames in medical images processed.