Reversible Data Hiding in Color Image Based on Channel Linear Regression and Two-dimensional Multiple Prediction Error Histogram Shifting

we propose to generate prediction error histograms using channel linear regression.Then, we embed secret bits using two-dimensional histogram shifting in the order of local space complexity of pixels.At the receiver side, similar processes are applied and additional data is extracted in reverse order to ensure reversibility.The experimental results show that the proposed method achieves better visual quality in terms of PSNR under the same embedding capacity compared to existing methods.