An Overview

An Overview & Analysis of Image
Steganography Techniques

Aashish Shrivastav1, Amit Tak1, Anirudh Chouhan1, Naman Jain1
1Department of Computer
Engineering & Information
Government Engineering
College, Ajmer, Rajasthan,
India [email protected]

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

Abstract: Steganography is becoming an important area of research in recent years. It is
defined as the study of invisible communication. It is an art and the science of embedding
information into cover image viz., text, video, audio or multimedia content for military
communication, authentication and many other purposes. It deals with the ways of hiding
the communication message and its existence from the unintended user. In image
steganography, secret communication is achieved through embedding a message into an
image as cover file and generates a stego-image having hidden information so, in short; we
can say that Steganography basically deals with the techniques of hiding the communication
data which is in existence in such a way that it remains confidential. There are several image
steganography techniques are used each have its pros and cons. This paper discusses
various image steganography techniques such as Least significant bit (LSB), Discrete
wavelet transformation, Pixel value differencing, Discrete cosine transformation (DCT),
Masking and filtering etc.

Keywords: Steganography, stego-image, Spatial domain methods, Transform domain
techniques, Distortion techniques, Cryptography.

1. Introduction

In today’s world, with the advancement of computer, communication and the rise of internet,
the information is easily transferred from one location to another. But in some cases, it is
needed to keep the information must travel secretly. One of the grounds discussed in
information security is the exchange of information through the cover signatures, spread
spectrum etc. that conceal the existence of information. But now day’s digital approaches
are used so the steganography is mostly used on digital data. Thus, steganography hides the
existence of data so that no one can detect its presence. In steganography the process of
hiding information content inside any multimedia content like image, audio, video is
referred as a “Embedding”. For increasing the confidentially of communicating data both
the techniques may be combined. In this paper, we describe brief review and analysis of
several image steganography techniques.

2. Image Steganography Techniques

Image Steganography is the process of hiding the sensitive information into the cover image
with no degradation of the image and providing better security so that unauthorized user cannot
access the hidden information. Hiding the data by taking the cover image is referred as image
steganography. In image steganography pixel intensities are used to hide the data. In digital
steganography, images are widely used cover source because there are number of bits present
in digital representation of an image.

Figure 1 shows the various image steganography techniques. Image steganography techniques
are broadly classified into following: –

Figure 1: Various Image Steganography Techniques.

Image steganography terminologies are as follows: –

• Cover-Image: Original image which is used as carrier for hidden information.

• Message: Actual information which is used to hide into images. Message could be plain
text or some other image.

• Stego-Image: after embedding message into cover image is known as stego-image.

• Stego-Key: A key is used for embedding or extracting the messages from cover-image
and stego-images.

Figure 2: Image Steganography

Generally, image steganography is categorized in following aspects 23 and Table-1 shows
the best steganographic measures: –
• High Capacity: Maximum size of information can be embedded into image.
• Perceptual Transparency: After hiding process into cover image, perceptual quality
will be degraded into stego-image as compared to cover image.
• Robustness: After embedding, data should stay intact if stego-image goes into some
transformation such as cropping, scaling, filtering and addition of noise.
• Temper Resistance: It should be difficult to alter the message once it has been
embedded into stego-image.
• Computation Complexity: How much expensive it is computationally for embedding
and extracting a hidden message?

Table 1. Image Steganography Algorithm Measures
Measures Advantage Disadvantage
High Capacity High Low
Perceptual Transparency High Low
Robustness High Low
Temper Resistance High Low
Computation Complexity Low High

2.1 Spatial Domain Methods

In spatial domain steganography method, for hiding the data some bits are directly changed in
the image pixel values. There are many versions of spatial steganography, all directly change
some bits in the image pixel values in hiding data. Least Significant Bit (LSB) – based
steganography is one of the simplest techniques that hides a secret message in the LSBs of
pixel values without many perceptible distortions. Changes in the value of the LSB are
imperceptible for human eyes. Most used method in this category is least significant bit Spatial
domain techniques are classified into following-

1. Least Significant Bit (LSB)
2. Pixel Value Differencing (PVD)
3. Edges Based Data Embedding Method (EBE)

4. Random Pixel Embedding Method (RPE)
5. Mapping Pixel to Hidden Data Method
6. Labelling or Connectivity Method
7. Pixel Intensity Based Method
8. Texture Based Method
9. Histogram Shifting Methods

General advantages of spatial domain LSB technique are:
1. There is less chance for degradation of the original image.
2. More information can be stored in an image.

Disadvantages of LSB technique are:
1. Less robust, the hidden data can be lost with image manipulation.
2. Hidden data can be easily destroyed by simple attacks.

2.1.1 Least Significant Bit (LSB)
LSB insertion is a common and simple approach for embedding information in a cover file.
Since LSB is replaced there is no effect on cover image and hence unintended user will not get
the idea that some message is hidden behind the image 3. However, a little change in level of
intensity of original and modified pixel, but it cannot be detected visually.

Digital images used as cover file are mainly of two types- 24-bit images and 8-bit images. In
24-bit images we can embed three bits of information in each pixel. In 8-bit images, one bit of
information can be hidden into images. After applying the LSB algorithm the image obtained
having secret message is called stego-image. LSB technique as the name implies replaces the
least significant bit of the pixel with the information to be hidden.

The following example explain how the letter A can be hidden into the three pixels i.e. eight
bytes of a 24-bit image.

Pixels: (00100111 11101011 11001010)
(00100111 11011000 10101001)
(11001000 00110111 11011001)

A: 010100111

Result: (00100110 11101011 11001010)
(00100111 11011000 10101000)
(11001001 00110111 11011001)

The main advantage of LSB method is easy to implement and high message payload and there
is less chance of degradation of quality of original image. The disadvantages are that the
information can be easily extracted or destroyed by simple attacks and it is less robust,
vulnerable to image manipulation.

2.1.2 Pixel Value Differencing (PVD)

In PVD method, gray scale image is used as a cover image with a long bit-stream as the secret
data 4. It was originally proposed to hide secret information into 256 gray valued images. The
method is based on the fact that human eyes can easily observe small changes in the smooth
areas but they cannot observe relatively larger changes at the edge areas in the images. PVD
uses the difference between the pixel and its neighbour to determine the number of embedded
bits. The larger the difference amount is, the more secret bits can be embedded into the cover
This method is proposed to enhance the embedding capacity without improper visual
changes in stego-image. But the disadvantage of the method is sometimes the pixel value in
the stego-image may exceed the range 0-255 which leads to improper visualization of the stego
image. It has also weak security performance due to non-adaptive quantization, embedding
some information in smooth areas etc
It scans the image starting from the upper left corner in a zigzag manner. Then, it simply
divides the cover image into number of blocks where each block consists of two consecutive
non-overlapping pixels. The difference of the two pixels in the block is used to categorize the
smoothness properties of the cover image. A small difference value indicates that the pixels are
at smooth area whereas pixels around edge area have large difference values. The data is
embedded mostly in the edge areas because the changes of the pixel values are more easily
noticed by human eyes. Therefore, in PVD method a range table has been designed with n
contiguous ranges Rk (Where k=1,2,3……n) where the range is between 0 to 255.The lower
and upper bound are denoted as Ik and Uk respectively, then Rk €Ik ,Uk. The width Wk and
Rk is calculated by WK= UkIk+1 which decides how many bits can be hidden in pixel block.
When extracting the embedded data from stego-image original range table is required.

2.1.3 Histogram Shifting Method

Histograms are used for graphical representation of image. It represents the pixel value and
density at a particular pixel. It plots the pixel for each part of the image. A histogram is useful
to identify pixel distribution, density of colours and tonal distribution. A histogram provides
the highest and lowest pixel values in graph. Histogram shifting is the technique which is used
to modify or to extract a certain group of pixels from a image 11. In histogram the highest
value is called maxima and the lowest value is called minima. When the pixel value is modified
for embedding process it should not cross the minima and maxima limit. There are several
algorithms which supports histogram functionality in order to manipulate the image. The
number of the pixels constituting the peak in the histogram of a cover image is equal to the
hiding capacity because a single peak in a cover image is used 5.
Several histograms shifting techniques are enhanced by dividing the cover image into
blocks to generate a respective peak for each block which provides more hiding capacity into
the multiple blocks.

2.2 Transformation Domain Technique

Transformation domain methods hides message in the significant areas of the cover image
which makes them more robust against various image processing operations like compression,
cropping and enhancement. There are many transformation domain methods exists. The basic

approach used for hiding information is to transform the cover image, tweak the coefficients
and then insert the transformation.

This is a more complex way of hiding information in an image. Various algorithms and
transformation are used on the image to hide information in it. Transform domain embedding
can be termed as a domain of embedding techniques for which a number of algorithms have
been suggested 17. The process of embedding data in the frequency domain of a signal is
much stronger than embedding principles that operate in the time domain. Most of the strong
steganographic systems today operate within the transform domain techniques have an
advantage over spatial domain techniques as they hide information in areas of the image that
are dependent on the image format and they may outrun lossless and lossy format conversions.

Transform domain techniques are broadly classified into: –
1. Discrete Fourier Transformation Technique (DFT)
2. Discrete Cosine Transformation Technique (DCT)
3. Discrete Wavelet Transformation Technique (DWT)
4. Lossless or reversible Method
5. Embedding in coefficient Bits

2.2.1 Discrete Fourier Transformation (DFT) Technique

In DFT all the insertion of hidden message is done in the frequency domain. It is a more
complex way of hiding message into frequency domain of the image. The Discrete Fourier
Transform of spatial value f (x, y) for an image of size M × N is defined in equation for
frequency domain transformation 6.


Similarly, inverse discrete Fourier transform (IDFT) is used to convert frequency component
of each pixel value to the spatial domain value and the equation for transformation from
frequency to spatial domain is


When DFT is applied it converts the cover image from spatial domain to frequency domain
and each pixel in spatial domain is transformed into two parts: real and imaginary part. The
hidden message bits are inserted in real part of frequency domain excluding first pixel. After
embedding IDFT is performed frequency domain converted into spatial domain. During the
extraction or decoding of the message image from spatial domain is transformed to frequency
domain. After applying DFT and extraction algorithm the original source image is retrieved.

2.2.2 Discrete Cosine Transformation (DCT) Technique

The DCT transforms the image from spatial to frequency domain and separates the image into
spectral sub-bands with respect to visual quality of the image, i.e. low, middle and high
frequency components as shown in fig. 2. Here FL and FH are used to denote the lowest
frequency components and higher frequency components respectively. FM is used as
embedding region to provide additional resistance to lossy compression techniques, while
avoiding significant modification of the cover image 13.

Figure 2: DCT Regions

It is used in the JPEG compression algorithm to transform successive 8 × 8-pixel blocks of the
image into 64 DCT coefficients each in frequency domain. Each DCT coefficient F(u, v) of an
8 × 8 pixel block of image pixels f(x, y) is
calculated by


Where C(x)=1/ when x=0 and C(x)=1 otherwise. The following quantization operation is
performed after calculating the coefficients:


Where Q(u,v) is a 64-element quantization table. The hidden message is embedded into the
redundant bits, i.e. the least significant bits of the quantized DCT coefficients. A modification
of a single DCT coefficient affects all 64 image pixels. In DCT based techniques, the secret
data is embedded in the carrier image for DCT coefficients lower than the threshold value 7.
Pixels having DCT coefficient value below threshold are known as potential pixels. Hence to
avoid the visual distortion in image the potential pixels are used for data hiding.

2.2.3 Discrete Wavelet Transformation (DWT) Technique

The Discrete Wavelet Transformation Technique is the new idea in the applications of the
wavelets. The standard technique of storing in the least significant bit of pixel still applies but
the only difference is the information is stored into the wavelet coefficients, instead of changing
the bits of actual pixels in the image. DWT have advantage over Fourier Transformation, it

performs local analysis and multi-resolution analysis. Wavelet analysis can reveal signal
aspects like discontinuities, breakdown points etc. more clearly than Fourier Transformation.

The DWT splits the signal into two parts- high and low frequency. The information about the
edge component is in high frequency part and the low frequency part is further split again into
high and low frequency parts. A one-dimensional DWT uses filter bank algorithm 12 and the
information is convolved with high pass filter and low pass filter. Human eyes are less sensitive
to high frequency so high frequency components are used for steganography.

In two dimensional applications, for each level of decompositions, we first perform the
DWT in the vertical direction, followed by DWT in the horizontal direction 8. As we can see
in the fig.3, the first level of decomposition results into four classes or sub-band: approximate
band(LL1), vertical band(LH1), horizontal band(HL1), diagonal detail band(HH1). The
approximation band consists of low frequency wavelet coefficients which contains the
significant part of the spatial domain image. The other bands consist of high frequency
coefficients, which contain the edge details of the spatial domain image. For each successive
level of decomposition, the approximate band of the previous level is used as the input. In
second level of decomposition, the DWT is applied on LL1 band which decomposes it into
four sub-bands: LL2, LH2, HL2 and HH2.

Figure 3: Three phase decomposition using DWT

2.3 Distortion Technique

In distortion techniques the information is stored by signal distortion. These techniques require
the knowledge of the original cover image during the decoding process. The encoder applies
series of modifications to the cover image and the decoder functions to check for the various
differences between the original cover image and distorted cover image to recover the secret
message. Using this technique, a stego object is created by the sender by applying a sequence
changes to the cover image. This sequence of modification corresponds to a specific secret
message required to transmit. The message is encoded at pseudo- randomly chosen pixels in
the image. If the stego-image differ from the cover image at the given message pixel, the
message bit is a “1” otherwise “0”. The sender can modify the “1” value pixels in such a way
that the statistical properties of the image should not affected.

Distortion techniques need knowledge of the original cover image during
the decoding process where the decoder functions to check for differences between the original
cover image and the distorted cover image in order to restore the secret message. The encoder

adds a sequence of changes to the cover image. So, information is described as being stored by
signal distortion 18.

Using this technique, a stego object is created by applying a sequence of modifications to
the cover image. This sequence of modifications is use to match the secret message required
to transmit 19. The message is encoded at pseudo-randomly chosen pixels. If the stego-image
is different from the cover image at the given message pixel, the message bit is a ‘1’, otherwise,
the message bit is a ‘0’. The encoder can modify the ‘1’value pixels in such a manner that the
benefits of this technique.

In any steganographic techniques, the cover image should never be used more than once.
If an attacker tampers with the stego-image by cropping, scaling or rotating, the receiver can
easily detect it. In some cases, if the message is encoded with error correcting information, the
change can be reversed and the original message can be recovered 20.

The receiver must have access to the original cover for retrieving the message; it limits
the benefits of this technique. In every steganography techniques, the cover image should never
be used more than once. If an attacker has access to the cover image the secret message can be
easily detected by attacker from the stego-image by cropping, scaling or rotating it. In some
cases, if the message is encoded with error correcting information, the change can even be
reversed and the original message can be recovered 9.

2.4 Masking and Filtering

This technique is usually applied on 24 bits or grayscale images, uses a different approach to
hiding a message. It hides information by marking an image, similar to paper watermarks. This
technique actually extends an image data by masking the secret data over the original data as
opposed to hiding information inside of the data 10. These techniques embed the information
in the more significant areas of the image than just hiding it into noise level. Watermarking
techniques can be applied on the image without the fear of its destruction due to lossy
compression as they are more integrated into the image.

This method is more robust than LSB modification with respect to compression and
different kinds of image processing since the information is hidden into the visible parts of the
image. The main drawback of this technique is that it can only be used on gray scale images
and restricted to 24-bit images.

These techniques hide information by marking an image, in the same way as to paper
watermarks. These techniques hide the information in the significant areas than just hiding it
into the noise level. The hidden message is more integral to the cover image. Watermarking
techniques can be applied without the fear of image destruction due to lossy compression as
they are more integrated into the images.

Advantages of Masking and Filtering Techniques: –
• This method is much more robust than LSB replacement with respect to compression
since the information is hidden in the visible parts of the image.

Disadvantages of Masking and Filtering Techniques: –
• Techniques can be applied only to gray scale images and restricted to 24 bits.

3. Analysis of Steganography Techniques

Table 2. Comparison of Steganography Techniques 22

Domain Algorithm Invisibility Capacity Robustness Security Complexity
LSB High 1-3 bpp Low Low Low
PIT Medium >1 bpp Low High Low
OPAP Medium 1 bpp Low High Medium
na et al. 23
Very High 2 bpp 23 High High Medium
Ratnakirti Roy
et al. 26
Very High NA* High High Medium
JSteg Medium