Analysis of Reversible Watermarking using Image Compression and Histogram Shifting

Abstract

In this paper, we propose new reversible watermarking technique which is for data hiding. This data hiding enables embedding of messages in a host image without any loss of host content, which is proposed for image authentication that if the watermarked image is deemed authentic, we can revert it to the exact copy of the original image before the embedding occurred. Our first contribution is towards the modulation of the Histogram Shifting on image contents with compression of image. we present an improved histogram-based reversible data hiding technique based on pixel prediction and sorting. A rhombus pixel prediction is employed to explore the prediction for histogram-based embedding. Sorting the prediction has a good effect on increasing the embedding capacity. The second contribution is to hide data into textured area of the image. To achieve large hiding capacity we used characteristics of the pixel difference by keeping low distortion. The proposed scheme which derives a two-stage embedding strategy to solve the problem about communicating peak points. We also present a histogram shifting technique to prevent overflow and underflow. In this way, the watermark embedder and extractor synchronized for message extraction and image reconstruction.