Embodiments of the invention are a novel splicing detection scheme that detects the spliced images by distinguishing image features that are extracted by exploiting both magnitude and phase information of a given image. The image features include the statistical moments of characteristic functions of wavelet subbands of a test image and a prediction-error image. In addition, the approximation (LL) subband at different levels is individually erased by forcing the wavelet coefficients to zero and the inverse wavelet transform is applied in order to produce reconstructed image with enhanced high frequency components. Further, the moments of the characteristic functions of these reconstructed images provide additional image features. Moreover, the statistics (mean, variance, skewness and kurtosis) of 2-D phase congruency array associated with the reconstructed images provide additional image features for splicing detection. These inputs provide a 120 dimensional image feature vector that includes 96 moment features and 24 phase features.