Data insertion and scrambling for unified scalable information hiding / Ong Sim Ying
Information hiding is generally divided into two disciplines, namely, external data insertion and scrambling. External data insertion embeds data into the multimedia content for various purposes, including enrichment (i.e., metadata and hyperlink), owner identification (i.e., fingerprint), proof...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/5519/1/CD_Cover.pdf http://studentsrepo.um.edu.my/5519/2/WHA100032_ONG_SIM_YING_PhD_Thesis.pdf http://studentsrepo.um.edu.my/5519/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Information hiding is generally divided into two disciplines, namely, external data
insertion and scrambling. External data insertion embeds data into the multimedia content
for various purposes, including enrichment (i.e., metadata and hyperlink), owner identification
(i.e., fingerprint), proof of ownership (i.e., watermark), covert communication
(i.e., steganography), etc., depending on the applications in question. On the other hand,
scrambling conceals the perceptual meaning of the content, either completely or partially,
depending on the requirement of the intended application.
Conventionally, these two disciplines are studied independently. The basic properties
of these two disciplines include reversibility, embedding capacity and commutative.
However, due to the massive use of digital contents in our daily life, unification of these
two disciplines (hereinafter referred to as unified information hiding) is needed to cope
with multiple requirements in a single application.
In this thesis, the background of information hiding, research motivations, objectives
and methodology are presented in Chapter 1. Next, Chapter 2 proposes the classification
of unified information hiding and reviews the relevant state-of-the-art methods. The
following chapters propose four different unified information hiding approaches in the
spatial and compressed domains for image.
Specifically, in Chapter 3, an Embed-To-Scramble method, namely, Histogram Association
Mapping (HAM), is proposed to overcome the problems in conventional histogram
manipulation method in the spatial domain. An image is first divided into nonoverlapping
blocks, and each block is further classified into two categories to perform
their respective designated scrambling-embedding operations. This method is reversible,
scalable and the embedding capacity is relatively high when compared to the conventional
methods.
iii
Chapter 4 puts forward a scalable Scramble-Then-Embed method named SURIH to
realize scrambling-embedding in the Discrete Cosine Transform (DCT) domain. Adaptive
scanning order is proposed to suppress the bitstream size increment caused by encoding
the side information of embedding technique. Unlike the conventional methods that
confine their scope of permutation to a block, SURIH permutes the coefficients globally
throughout the entire image. Performance of SURIH is evaluated at the end of the chapter.
In Chapter 5, two Scramble-To-Embed methods are proposed in the spatial and
DCT compressed domain, in which the recovery processes rely on the natural properties
of the utilized components. The utilized components in the DCT compressed methods
include DC coefficient, AC Block Energy (i.e., sum of magnitude of each 8×8 block),
and ZRV (zero-run, AC coefficient value) pairs. Reversibility is achieved by exploiting
the natural property of each component, along with the aid of minimum amount of side
information. Similar techniques are also applied in the spatial domain, and it is further
discussed in this chapter. At the end of the chapter, experimental results are recorded and
compared with the conventional methods.
Chapter 6 discusses the important properties achieved in each proposed method and
their limitations in detail. Specifically, the embedding capacity and image quality can be
gradually controlled by tuning the parameters introduced in each proposed method. The
reversibility and commutative property of each method are also discussed.
Finally, Chapter 7 concludes this thesis and discusses the future work of scalable
unified information hiding. |
|---|