Application of wavelet and particle swarm optimization in steganography

An implementation of particle swarm optimization (PSO) for image steganography is presented. Message image of 256×256 is decomposed by using Daubauchi-1 wavelet to five levels. Only approximation matrix is considered in all levels of decompositions. At the fifth level of decomposition, the approximation matrix size is 8×8. The information in the approximation matrix is hidden in the lower nibble of the cover image using the particles locations obtained by training the PSO algorithm. The number of elements of the 8×8 matrix is 64. Hence, a minimum of 64 particles is generated. Based on a specific generation number and the current best value, the locations of the particles are stored. In the information hiding process, each coefficient value of the 8×8 matrix is stored in the corresponding locations in the cover image using least significant bit process (LSB).