Defining redundancy in Histogram of any signal for reversible data embedding

Conventionally, sparse-histogram consists of redundant vacant bins that are is utilized for data embedding in histogram mapping based methods. However, when all bins are occupied, i.e., the histogram is not sparse, these methods are not applicable reversibly. In this paper, we propose a method that utilizes the universal parser to model any signal such that redundancy (i.e., sparse-histogram) can certainly be defined regardless its underlying statistical features of the signal. First, the universal parser recursively partitions the bitstream of signal into tuples until a sparse-histogram is obtained. Next, the redundancy in sparse-histogram is utilized to achieve data embedding by a proposed method, which associates tuples that occur in histogram with tuples of count zero. After that, to embed “1” from payload, a tuple is mapped to its associated counterpart tuple. However, no mapping is carried out to embed “0”. Empirically, it is verified that the proposed method is capable to define sparse-histogram (i.e., venues for data embedding) in any signal. The proposed method achieves average carrier capacity of ~ 31732 and ~ 126 bits in images of sparse and non sparse-histogram, respectively. Also, the proposed method is file-size preserving for images of sparse-histogram and causes limited file-size increase by 0.9% in images of non sparse-histogram.