02993ntm a2200301 i 4500003000800000005001700008006001900025007000300044008004100047020003300088040002700121100003000148245011200178264003600290264001200326300005900338336002100397337002500418338002300443347002400466500002500490502007300515504004000588520195200628610004002580650004502620650002602665MY-KuUP20251125111039.0t||||fr|||| 000 0 ta250514t20242024my a|||fr|||| 000 0 eng d  aTHE0010059 (Local)qHardback  aUMPSAbengcUMPSAerda0 aAfrig Aminuddin,eauthor.10aImage authentication and self-recovery with texture preservation in fragile watermarking /cAfrig Aminuddin 1aKuantan, Pahang :bUMPSA,c2024 4c© 2024  axiII, 130 pages :billustrations ;c30 cm. +e1 CD-ROM  2rdacontentatext  2rdamediaaunmediated  2rdacarrieravolume  2rdaatext filebPDF  aFaculty of Computing  aThesis (Doctor of Philosophy) -- Universiti Malaysia Pahang – 2024  aIncludes bibliographical references3 aThis research proposes a novel image watermarking scheme with self-recovery capability  to authenticate and recover tampered images using the embedded watermark data. The  scheme employs a new texture preservation technique to maintain the details of the image  texture in the recovered image. The texture set is adaptively computed according to the  image content. Each image has its own texture set tailored to the characteristics of the  image. The scheme generates the watermark data from each image block, comprising average, range, and texture values. The scheme embeds the watermark data into 2 LSB of the original cover image using a block map and a secret key. The scheme also employs  a texture preservation technique to enhance the quality of the recovered image. The  scheme utilizes a three-layer authentication process to detect and localize the tampered  regions. The scheme solves the issue of tampering coincidence using a new block  mapping technique, a tamper coincidence localization, and an image inpainting  technique. The experimental results show that the proposed scheme can achieve highquality watermarked and recovered images and high tamper detection and localization  accuracy under various attack scenarios. The texture preservation technique produces a  high-quality recovered image with a PSNR value of 40.04 dB and SSIM value of 0.9953  under 10% tampering rates on the USC-SIPI dataset. The scheme also outperforms  several existing watermarking schemes with self-recovery regarding the imperceptibility of the watermarked and recovered images. Image authentication is crucial for legal  validity, ensuring digital evidence is admissible and credible in court. In journalism, it  prevents the spread of misinformation, maintaining the integrity of news visuals. Across  various sectors, including healthcare and historical archives, it upholds accuracy and  trustworthiness in using and preserving digital images.20aFaculty of ComputingxDissertations 0aUniversities and collegesxDissertations 0aThesesxDissertations