Analysis of Frequency Domain Image Steganography using DCT and DWT

Abstract

Steganography is a way of hiding messages during communication within some cover media. Image Steganography the process of hiding secret messages in a cover image. Frequency domain hides the secret bits of input message within the significant parts of the cover image. It encodes bits of the message into the transform domain coefficients of the image. Discrete Cosine Transformation (DCT) and Discrete Wavelet Transformation (DWT) come under frequency domain technique. This research evaluates the DCT, DWT and a hybrid DCT with DWT algorithms using four measures Peak Signal to Noise Ratio (PSNR), Mean Square Error (MSE), Number of Pixel Change Rate (NPCR), and histogram. The result shows that DCT with DWT is better than the two algorithms DCT and DWT. The PSNR value of the combined method has the highest value of the other two techniques. DCT with DWT preserves the originality of input image and the stego image without losing image properties. The average PSNR value for DCT with DWT is 77.59, DWT is 71.07 and DCT is 73.08 and it signifies DCT with DWT is better. The average encoding and decoding time for DCT with DWT is 0.15, DWT is 0.43 and DCT is 0.46. Thus DCT with DWT being feasible for computational point of view. The differential analysis values for DCT with DWT, DWT and DCT with an average of NPCR are 92.60%, 91.56% and 91.43% respectively. All of the three algorithms are found to be good on histogram analysis producing similar histogram distributions of input and stego images.