Description

< The identification of input images into original or stego based on the features extracted from the images has numerous applications. The input images may contain any of the secret informations in them. For the identification of the original and the stego images extraction of the features from images were done based on the three domains of the images like spatial domain, spectral domain. DCT and DWT were employed for the conversion of the input images from the spatial and spectral domain correspondingly. The DCT transforms the images to frequency domains. A discrete cosine transform expresses a finite sequence of data points in terms of a sum of cosine functions oscillating at different frequencies. The cosine transformation splits the image into high and low coefficients. The informations in the high coefficient is not upto the visual perception of the human eyes. The required informations were present in the low coefficient contents of the image. The wavelet transformation converts the images into four coefficients like LL, LH, HL and HH. The LL coefficients contains the pixel information while the other coefficients contains the spectra informations. The statistical parameters like mean, standard deviation, Energy, Entropy, Correlation coefficients and histogram were calculated from the signals and the extracted values were used as features. The extracted features were then classified using SVM and Neural Network Classifier. The classification process in SVM is based on the kernel functions which were identified for the identification of the classes in the images. The neural network identifies the categories based on the generated weightages for the neurons that are generated in the layers of the network. The classification performance is measured based on the confusion matrix.