A novel triple recovery information embedding approach for self-embedded digital image watermarking


Gul E., Öztürk S.

MULTIMEDIA TOOLS AND APPLICATIONS, vol.79, pp.31239-31264, 2020 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 79
  • Publication Date: 2020
  • Doi Number: 10.1007/s11042-020-09548-4
  • Journal Name: MULTIMEDIA TOOLS AND APPLICATIONS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, FRANCIS, ABI/INFORM, Applied Science & Technology Source, Compendex, Computer & Applied Sciences, INSPEC, zbMATH
  • Page Numbers: pp.31239-31264
  • Erciyes University Affiliated: Yes

Abstract

Image tamper detection and recovery has become an important issue in recent years. In order to detect and recover high tampering rated images, this paper presents a novel self-embedded fragile watermarking method with triple recovery information embedding approach. In this proposed method host image is divided into sixteen main blocks. Four partner blocks are selected from main blocks as a partner group using Look-Up table which is constructed for recovery against high tampering rated images. Each partner block is divided into 4 x 4 blocks. Recovery data of each partner block is generated from the mean values of the 4 x 4 blocks. Then, triple recovery information for each partner block is constructed by combining recovery data of the three other partner blocks. After the generation of the recovery information, each partner block is divided into 16 x 16 blocks. Then, these blocks are divided into four 8 x 8 sub-blocks. Recovery bits for each partner block are embedded into the first and second least significant bits (LSBs) of the first three sub-blocks. It means that triple copies of recovery information for each partner block are embedded into three other blocks in the image, which provides triple chance for recovery of the tampered areas. On the other hand, authentication information is generated from the pixels of the 16 x 16 recovery information embedded image blocks by using MD5 hash function. This authentication information is embedded into the quarter part of each 16 x 16 block by using LSB substitution. The performance of the proposed method has been demonstrated by applying different size of cropping attacks to the different areas of the watermarked images. Experimental results show that the proposed method satisfactory recovers up to 75% tampering rated images.