A reversible data embedding approach based on inter-channel gradient shifted MSB predictor in encrypted images for cloud applications

Abstract

The paper proposes an encrypted image-based reversible data embedding approach using an inter-channel gradient-shifted MSB predictor for the application of cloud storage. In this approach, the data embedding was done on the cloud to store the encrypted images. Initially, the image was encrypted using permutation-based encryption by the user which is uploaded to the cloud. From the encrypted images, a primary channel and two secondary channels are estimated from which the gradient images are estimated. Using the histogram, the gradient images are estimated which is then shifted to perform embedding. Three different types of shifting approaches are proposed which include minimum value gradient shifting, threshold value gradient shifting, and maximum correlated gradient shifting (MC-GS). The gradient-shifted images are used to embed the data using the MSB predictor approach. The analysis of the algorithm was done using the standard color images obtained from the SIPI dataset and the evaluation was done with measures such as structural similarity index (SSI), peak signal-to-noise ratio, embedding rate, and entropy. The MC-GS gradient shifting results in an SSI, PSNR, and embedding rate of 0.1046, 8.13 dB, and 2.832 bpp respectively.