A comparative study of frequency domain based approaches for image tamper detection

Detection of image tampering is required to distinguish forged images from authentic images. In this paper, frequency domain transformation techniques such as Discrete Wavelet Transform (DWT) and Discrete Cosine Transform (DCT) have been used as a means to detect image tampering. These techniques have been applied in various combinations. Principal Component Analysis (PCA) is applied to frequency coefficients to further reduce their dimensionality. Classification is done using Least Square Support Vector Machine (LS-SVM). Experimental results show that use of DWT and DCT along with a sliding window of size 2 provides best accuracy in detection of tampered images for CASIA dataset.