Data Hiding on Image by Using Particle Swarm Optimization and Histogram Modification

Steganography is used to hide a secret message within a cover image, thereby yielding a stegoimage such that even the trace of the presence of secret information is wiped out. This project proposes a novel prediction based reversible steganographic scheme based on image Inpainting. First, reference pixels are chosen by using Particle Swarm Optimization technique. Then, In order to generate a prediction image, the image Inpainting technique based on partial differential equations is introduced, which has similar structural and geometric information as the cover image. Finally, by using the two selected groups of peak points and zero points, the histogram of the prediction error is shifted to embed the secret bits reversibly. Since the same reference pixels can be exploited in the extraction procedure, the embedded secret bits can be extracted from the stego image correctly, and the cover image can be restored lossless. Through the use of the PSO algorithm and the inpainting predictor, the prediction accuracy is high, and more embeddable pixels are acquired. Thus, this project provides a greater embedding rate and better visual quality compared with recently reported methods.