An approach for anti-forensic contrast enhancement detection using grey level co-occurrence matrix and Zernike moments

An approach for anti-forensic contrast enhancement detection using grey level co-occurrence matrix and Zernike moments

연구 분야: Strategies

학회: International Journal of Information Technology

The present paper aims to detect contrast enhancement when forensic fingerprints are removed by anti-forensic attacks. The methodology employed in this study exploits statistical anomalies in the frequency domain by using second-order statistics determined from grey-level co-occurrence matrix (GLCM). The magnitude of first 36 Zernike moments (ZMs) of column-wise Fourier transform of the GLCM is used to generate the feature vector. A support vector machine (SVM) classifier is employed to distinguish between original and altered images. To evaluate the performance of presented model, we plot receiver operating characteristic (ROC) curve and calculate true positive rate (TPR), false positive rate (FPR), and accuracy of the model. The results show that in the presence of an anti-forensic attack, the TPR reaches 92.0%, and the FPR reaches 91.1%. Thus, the results verify the effectiveness of the proposed approach for detecting contrast enhancement when anti-forensic attacks are removing forensic fingerprints. The proposed method is also robust against Gaussian white noise and losses due to compression.