A Blind Reversible Data Hiding Method for High Dynamic Range Images Taking Advantage of Sparse Histogram

Abstract

This paper proposes a method of reversible data hiding (RDH) for high dynamic range (HDR) images. An RDH method once distorts an image to hide data to the image, and the method takes data out and simultaneously recovers the original image without any distortion from the distorted image carrying hidden data. Whereas conventional RDH methods are for ordinary images whose pixel values are uniformly quantized integers, the proposed method focuses HDR images whose pixel values are non-uniformly quantized floating-point numbers. HDR images have a sparse histogram, i.e., many zero points are scattered over the tonal distribution of images, and the method modifies multiple peak and zero points of the histogram to hide data to an image. In addition, while an RDH method generally needs to memorize a set of image-dependent parameters for hidden data extraction and original image recovery, the proposed method is free from parameter memorization by introducing two mechanisms; restriction of histogram modification and a parameter hiding prior to data hiding. Moreover, keys are required to take hidden data out in the proposed method. Experimental results show the effectiveness of the proposed method.