Page 26 - 2023-Vol19-Issue2

22 | Taheri

to prevent removing the watermark image. DCT plays the role
of transparency of watermark in our algorithm. The robust-
ness of the algorithms can be improved through the extracted
watermark on post-processing that using the associative mem-
ory function of neural network. Implementation results show
that the Barbara image with more edge attribute has better ro-
bustness versus compression attacks. The Baboon image with
more texture features has more robustness versus noise attack.
The Cameraman image with less edge and texture attributes
has more robustness versus filtering and compression attacks.
IWT-DCT combination has more efficient than DWT-DCT
combination, this is the major fact, we can conclude from im-
plementation results, compare to our previous methods. Our
final conclusion is, we must find more efficient wavelet and
neural network to overcome our algorithm weaknesses.

Fig. 7. Watermarked Barbara Image

TABLE II.
IMPLEMENTATION RESULTS AND COMPARISONS FOR

BABOON IMAGE

Kinds of Attack on Barbara image Proposed Method Method in [13] Method in [14]
PSNR SIM PSN SIM PSNR SIM
Gaussian Noise 32.0 93.8 27.5 93.7 35.3 94
Low Pass Filter 36.2 90.5 25.0 89.0 35.3 93.9
Median Pass Filter 38.2 95.6 28.8 95.2 32.9 97.5
19.8 74.6 18.4 83.0 20.7 82.4
Scaling 1/5 44.6 96.0 37.2 97.9 40.5 97
Jpeg 75% 39.0 92.2 32.1 91.7 38.8 94.5
Jpeg 25% 31.3 90.8 26.1 88.3 33.4 91.7
Jpeg 10% 28.1 84.9 19.2 85.1 27.1 89.6
Jpeg 2000 with bitrate 3

Fig. 6. Original Barbara Image CONFLICT OF INTEREST

TABLE I. The author have no conflict of relevant interest to this article.
IMPLEMENTATION RESULTS AND COMPARISONS FOR
REFERENCES
BARBARA IMAGE
[1] A. Cheddad, J. Condell, K. Curran, and P. Kevitt, “Digi-
Kinds of Attack on Barbara image Proposed Method Method in [13] Method in [14] tal image steganography: Survey and analysis of current
PSNR SIM PSN SIM PSNR SIM methods,” International Journal of Signal Processing,
Gaussian Noise 32.9 93.8 27 98.3 31.3 92.3 Elsevier, vol. 90, no. 3, pp. 727–752, 2010.
Low Pass Filter 40.9 90.7 25.8 92.8
Median Pass Filter 50.0 96.9 27.4 95.7 29 88.8 [2] K. Vijay and D. Kumar, “Performance evaluation of dwt
24.2 77.9 20.0 86.3 34.7 97.6 based image watermarking,” IEEE 2nd International
Scaling 1/5 52.0 100 37.4 98.9 17.5 81.0 Advance Computing Conference (IACC), pp. 223–228,
Jpeg 75% 43.5 95.5 34.7 94.5 39 97.5 2010.
Jpeg 25% 35.4 90.6 28.2 91.2 36.9 93.8
Jpeg 10% 27.6 86.1 19.4 88.1 32.6 89.9 [3] H. Ahmed, “Hiding color image within a video file
Jpeg 2000 with bitrate 3 25.1 88.7 using discrete wavelets transform and artificial neural
networks,” 2nd International Conference on Electrical,