Watermarking of single and multi frame ultrasound medical images

• Main Author: Gran, Badshah
• Format: Thesis
• Language: English
• Published: 2016
• Subjects: Q Science (General); T Technology (General)
• Online Access: http://umpir.ump.edu.my/id/eprint/15257/19/Watermarking%20of%20single%20and%20multi%20frame%20ultrasound%20medical%20images.pdf; http://umpir.ump.edu.my/id/eprint/15257/

Digital communication of medical images may cause image manipulation for non-mandatory purposes. The originality of single and multi-frame medical images is strictly required to be used for patient health recovery. Watermarking is one of the latest security techniques, which can be used for protection of digital images. This dissertation has focused on watermarking of grayscale eight bits single and multi-frame ultrasound medical images. Watermarking of a medical image with heavy payload causes image degradation and directly affects patient diagnosis. Therefore, watermark must be lossless compressed to reduce its size without data loss. In the past Lempel-Ziv Welch (LZW) has been used for text, image pixel data lossless compression and in this research we extend this technique to binary watermark lossless compression to enhance the performance of region based watermarking techniques. In region based watermarking, an image is divided into region of interest (ROI) and region of non-interest (RONI). LZW compressed watermark based, a new region based watermarking scheme, tamper detection and recovery watermarking of medical images (TDARWMI) is developed. Produced results of the scheme are compared with those of other watermarking schemes and are found better in compression ratio, PSNR of watermarked and recovered images. In region based watermarking schemes, the tamper detection and recovery is restricted to ROI, while rest of the image remains under security threats. Possible wrong selection of ROI is another limitation of region based watermarking schemes. Non-proper selection of ROI may create problems during recovery at destination and can affect the required analysis. Therefore, another watermarking scheme, whole image authentication, tampers detection, localization and complete recovery (WIATDACR) is developed to provide security to a complete image. The LZW compressed watermark is embedded into image’s ROI and RONI parts randomly. At destination, watermark is extracted and used for image authentication, tampers detection, localization and lossless recovery. This scheme can locate tampers occurred at any part of image and can perform recovery without data loss. The signature based security technique of DICOM multi-frame images has deficiency in tamper detection and recovery. Multi-frame image authentication, tamper detection, localization and recovery (MFATDLAR) watermarking scheme is formulated to facilitate a DICOM image’s frame-wise authentication, tamper detection and lossless recovery. In this scheme the watermark of every frame is prepared, compressed and encapsulated into respective frame’s RONI part. At destination the extracted watermark is used for the frame’s authentication, tamper detection, localization and lossless recovery.