A Novel Forgery Detection Algorithm Based on Mantissa Distribution in Digital Images

Nowadays, digital image forgery detection is one of important topics in research world. In this paper, we propose a novel forgery detection algorithm using the logarithmic basis of Benford's law which states the mantissa of the logarithm of all practical numbers should be uniformly distributed. Based on this fact, the proposed method uses extracted features from mantissa distribution of discrete cosine transform (DCT) coefficients in JPEG images. Support vector machine (SVM) is used for classification to detect authentic and forged images based on these features. Results show that our proposed algorithm has the highest mean accuracy (99.78%), sensitivity (99.77%) and specificity (99.79%) in comparison with previous works on CASIA V1.0 dataset.