A Steganography Scheme on

JPEG Compressed Cover

1. Aim
2. Introduction
3. Differences between cryptography
and steganography
4. Steganography techniques
5. Desired features of steganography
6. Related works
7. Overview of proposed model
8. Embedding process
9. Extraction process
AIM:

To implement a stegenography
scheme on JPEG compressed image
having high embedding capacity
Introduction:
• The Internet is one of the most popular and the easiest medium for
transmission of digital data among people.
• Anybody can access these data and Internet itself does not provide
any protection on these data.
• To protect the confidentiality of the data, in general, two approaches
are mainly used, i.e., cryptography and steganography
Difference between Cryptography and
Steganography:

Cryptography:
• In cryptography, the encryption process transforms the secret data, i.e.,
known as plain text into cipher text using an encryption key.
• Its main concern is that the illegitimate users should not able to decrypt
the cipher text into plain text without knowing the decryption key.
• However, in cryptography, the cipher text resembles into an unreadable
form, so it attracts the opponent to exploit the content of the cipher text
by employing some cryptographic attacks.
Difference between Cryptography and
Steganography:
Steganography:
• Steganos means covered or secret and graphic means writing or drawing.
• Its objective is to hide the secret data into some other unsuspected cover
media so that the secret data will be visually imperceptible.
• Steganography hides the secret message within cover media, where the
cover media may be audio, video, text or an image
• The common intention of any steganography technique is to embed the
maximum number of secret message bits into the cover media in such a
way that both the cover media and the stego media do not differ in large
extent against human visual perception
Steganography techniques:

The secret data embedding into image can be realized into two domains, i.e., spatial
domain and transform domain.
1.Spatial domain:
• The secret data are embedded into the cover image by directly modifying each pixel
value of the cover image itself.
• In spatial domain based steganography schemes, the secret message are hidden into the
uncompressed cover image.
• The most common and simplest steganography approach in the spatial domain is the
Least Significant Bit (LSB) substitution method
• The spatial domain based steganography scheme performs superior to the transform
domain based steganography in terms of embedding capacity
• Ex: LSB, LPAP, OPAP etc.
Steganography techniques:

2. Transform domain :
• Nowadays, before transmission of the data over the communication
channel, generally, the compressed version of the data set are
preferred for efficient communication and storage.
• In the transform domain based steganography, the secret data can be
embedded into the compressed version of the cover image.
• Initially the cover image is transformed using any suitable
transformation tools like DCT, SVD, DWT etc., and subsequently the
secret data are embedded after modifying the selective transformed
coefficients.
Desired features of Steganography:
• high payload capacity

• high imperceptibility
This is measured using a parameter named peak signal to noise ratio
(PSNR) and mean square error (MSE).

• more robustness
Related Works:
• Several steganography schemes based on the JPEG compressed image data,
are found in literature like JSteg , F5 , OutGuess , the proposed scheme by
Chang et al and the proposed scheme by Liu and Liao.
• The embedding capacity of JSteg is found low due to the presence of less
number of permissible carriers quantized DCT coefficients for transmitting or
embedding the secret message bits.
• To increase the number of nonzero quantized DCT coefficients, Chang et al
used a modified quantization table during JPEG based image compression.
• Moreover, Westfeld and Pfitzmann noticed that steganographic scheme that
modifies the coefficients sequentially using the LSB methods, does not
survive against some statistical steganalysis like chi-square family attacks
Related Works:
• Both the JSteg and the proposed scheme of Chang et al. are not suitable to resist
against chi-square family attacks since both the schemes used LSB substitution
method for embedding the secret message bits sequentially into the quantized
DCT coefficients.
• OutGuess steganography scheme embeds the secret message bits into quantized
DCT coefficients nonsequentially by replacing the LSB bits of the quantized DCT
coefficients.
• The F5 is able to resist chi-square attack and also has higher embedding capacity
than the JSteg steganography scheme. But it suffers from the shrinkage effect
• The proposed scheme of Liu and Liao is capable to resist against various
steganalysis attacks
Overview of Proposed
Method:
Embedding Process:
The schematic diagram of the embedding process is shown in Figure 3
where the embedding process is
carried out in three major phases.
Phase 1:
Initially the secret message is encrypted
by any standard data encryption
algorithm. The encrypted message
is further converted into
a binary bits sequence and
Represent the encrypted bit sequence
S into a set of numbers
such that, S ={S1,S2,.....,Sn},
where each number,
Si is consist of two bits sequence
i.e.B2i-1, B2i from S
Phase-2: DCT Transformation of Image

DCT Definition:The discrete cosine transform (DCT) represents an image as a sum of sinusoids of
varying magnitudes and frequencies.
 The DCT has the property that, for a typical image, most of the visually significant information
about the image is concentrated in just a few coefficients of the DCT. For this reason, the DCT is
often used in image compression applications. For example, the DCT is at the heart of the
international standard lossy image compression algorithm known as JPEG. 
Embedding Process:
Embedding Process:
• Quantization:
Embedding Process:
Extraction Process:
Results:
Results:
Results:
Conclusions:
• In this project, we have proposed a JPEG cover image based steganography
with high visual quality along with high embedding capacity stego image.
• To improve the embedding capacity, we have used modified quantization
table that increases the number of non-zero middle-band quantized DCT
coefficients of every 8×8 blocks and two bits of encrypted secret message bit
sequence are embedded into each permissible quantized DCT coefficients
• To maintain the visual quality of the reconstructed stego image, the encrypted
secret message bit sequence is embedded into the middleband quantized DCT
coefficients according to the modified zigzag scanning order.
• The proposed scheme provides high quality reconstructed stego- image from
the JPEG stego-image and is capable to withstand various statistical attacks.
THANK YOU