Denoising of Stego-images for different noise models

Steganography plays an important role in secret communication systems, however the main drawbacks in Stego-images are presence of noise and degradation during acquisition or transmission. In this paper different noise models, including additive and multiplicative types, are studied They include Gaussian noise, salt and pepper noise and speckle noise. The criterion of noise removal problem depends on noise type by which image is corrupting. Several types of linear and nonlinear filtering techniques are proposed and applied for noise removal Experiments show that Mean Square Error (MSE), Peak Signal to Noise Ratio (PSNR) and Correlation (COR), are the most suitable assessment parameters.