Page 54 - 2024-Vol20-Issue2

50 | Alobaidi & Mikhael

4) Comments on The Performance Metrics Partition the Use the Sum Weights in each 2D DCT of Convert block
It is important to note that RMSE, SSIM, and PSNR have lim- Cover Image into Adaptive Block and Chose the the chosen Coefficients to
itations. They do not capture all aspects of image quality, such Algorithm
as human visual perception, and may not always correlate well Blocks Minimum Block Binary
with subjective evaluations. Therefore, it’s recommended to
use these metrics in combination with other quality assess- Stegoimage 2D IDCT LSB
ment methods and consider the specific requirements and Embedding
characteristics of the application or task at hand.
Convert
III. PROPOSED TECHNIQUE Secret data
to Binary
Fig. 1 shows the proposed technique. The method initiates
by dividing the original image into distinct blocks that do not (a) Encoding Process
overlap. To determine the weights of each coefficient within
the Cosine domain, an adaptive algorithm described in D.is Partition the Extract the Convert block Convert Extracted Bits
applied individually to each block. The block with coefficients Stegoimage into Designated Coefficients to Decimal to Obtain
that exhibits lower total weights compared to the remaining DCT Then to the Secret Message
blocks is selected. These chosen coefficients are converted Blocks Block
into a binary representation called ”cover in binary.” The se- Binary
cret data, specifically its binary form, is then embedded in the
least significant bit (LSB) of the cover in binary. Afterward, (b) Decoding Process
the block is converted back to its decimal representation. To
obtain the stegoimage, 2D IDCT is applied. At the receiver Fig. 1. The proposed technique utilized in steganography
end, the recipient can extract the secret data through partition- system. the two modules of the system are shown.
ing the image into blocks. The block dimensions have to be
exactly as the ones utilized in encoding process. Also, the TABLE I.
index(es) of the chosen block(s) has(have) to be securely sent PSNRS FOR SPATIAL/ DCT BLOCKS/PROPOSED
to the receiver. TECHNIQUES FOR BOSSBASE DATABASE( PROPOSED
The encoding part of the algorithm of the proposed tech-
nique/Encoding Process is as follows: BLOCKS SIZE IS 4 × 4)

1. Inputs:Cover Image, Secret Message Image Index Spatial DCT Blocks Proposed
1 85.5 50.16 86.3
2. Divide the cover image into non-overlapping blocks 2 86.75 50.22 87.26
3 86.3 50.25 87.26
4 86.3 50.19 88.51
5 87.84 50.23 86.3
6 90.28 50.16 87.26
7 88.51 50.22 87.84
8 86.75 50.22 87.26
9 87.84 50.17 86.3
10 87.84 50.28 86.75

3. Get weights of Cosine and DHT coefficients for each • Structural Similarity Index (SSIM)
block by Applying D.
• Peak Signal-to-Noise Ratio (PSNR)
4. Sum the weights of Cosine coefficients in each block
• Human Visual Observation
5. Convert chosen block(s) to Binary
The proposed system is compared against three other tech-
6. Convert Message to Binary niques: the traditional Spatial LSB, energy-based DCT in-
sertion (which employs total block energy as the selection
7. Insert bits of the message in LSB of the chosen blocks parameter), and LSB insertion technique in [12].

8. Convert blocks to Decimal, and transform to Spatial IV. RESULTS
domain via 2
The evaluation of the proposed technique is implemented as
9. Output: Stegoimage shown in the following categories:

The performance of the proposed technique is evaluated using 1. BossBase 10 samples database (Break Our Stegano-
the following metrics: graphic System Base) [18].
The database consists of collection containing 10, 000
• Root Mean Squared Error (RMSE)