Sorting methods and adaptive thresholding for histogram based reversible data hiding

This paper presents a histogram based reversible data hiding (RDH) scheme, which divides image pixels into different cell frequency bands to sort them for data embedding. Data hiding is more efficient in lower cell frequency bands because it provides prediction that is more accurate. Using pixel existence probability for some pixels of ultra-low cell frequency band, another sorting is performed. Employing these two novel sorting methods, we determine smooth areas of an image more efficient than other schemes. The smoother area of the image is selected for data embedding, the less distortion of the marked image may be achieved. In another proposal, we introduce hiding intensity analysis to determine optimum prediction error to embed data. In comparison with methods that sequentially choose prediction error, this analysis results in better quality of the marked image. In effect, the proposed scheme increases the hiding capacity for a specific level of the distortion comparing to existent RDH algorithms. Experimental results confirm that the proposed algorithm outperforms state of the art ones.