Volume 26 , Issue 1 , June 2024 , Pages 17-30
Dlsoz Abdalkarim Rashid 1 ; Marwan B. Mohammed 2 ; Iqbal A. Baki Mohammed 3 ; Tara Nawzad Ahmad Al Attar 4
1 Department of Computer Science, College of Science, University of Sulaimani, Sulaimani, Iraq
2 Department of Computer Science, College of Science, University of Al-Nahrain, Baghdad, Iraq.
3 Department of Computer Technologies Engineering, Al-Turath University College, Baghdad, Iraq
4 Department of Computer Science, College of Science, University of Sulaimani, Sulaimani, Iraq.
The art of discovering concealed messages submerged in digital media using
steganography in a secured form is known as steganalysis. Steganography and
steganalysis have both gotten a lot of significance from law enforcement
agencies and the media. More specifically, universal steganalysis techniques
have grown in admiration since they perform regardless of the embedding
technique. This paper comparing between three machine learning (ML)
algorithms: Support Vector Machine (SVM), Naive Bays (NB), and KNearest Neighbor (KNN). These algorithms detect and classify whether
images have embedded data within them or not; the images used in the data set
have been referred to as “cover images” which means no data is embedded or
“stego images” which means data had been submerged. The experiment in this
study was run on a genuine dataset of categorized images. Each image in this
dataset contains two embedding rates: half embedding (HE) and full
embedding (FE). As a result, the comparison of the aforementioned
algorithms evaluations showed that SVM is more accurate than NB and
KNN algorithms when applied to HE images data, with an accuracy of 0.88 for
the 50% test sample and 0.90 for the 30% test sample. When the test sample
50% was 0.82 and the sample 30% was 0.86 the SVM algorithm
outperformed also. The development of more accurate and effective techniques
using machine learning algorithms with help in the improvement of the security
of digital communication which in turn will prevent the unauthorized
transfer of sensitive information.