Histogram Based Optimized Reversible Steganographic Scheme

Abstract

A novel prediction-based Reversible Steganographic scheme based on image inpainting can be optimized by choosing the reference pixels using optimization technique. Partial differential equations based on image inpainting are introduced to generate a prediction image, from the reference image that has similar structural and geometric information as the cover image. Then the histogram of the prediction error is shifted to embed the secret bits reversibly using two selected sets of peak points and zero points. From the stego image, the cover image can be restored losslessly after extracting the embedded bits correctly. Since the same reference pixels can be exploited in the extraction procedure, the embedded secret bits can be extracted. Through optimization of reference pixels selection and the inpainting predictor, the prediction accuracy is high, and more embeddable pixels are acquired with greater embedding rate and better visual quality, and reduced time required for the stego00 image generation.