A prediction error based histogram association and mapping technique for data embedment

Abstract

A histogram association and mapping based embedding policy implants a group of bits, called message chunk, in a block. The strategy first calculates a distribution of pixel values of each block in the grayscale, called histogram. Each gray value acts as a bin of the histogram. Though the histogram makes up of 256 bins, only several bins hold these pixel values of a block. The implantation rules shift the pixel value containing bins as a single object to a new position in the histogram. The number of implanted bits is dependent on the range of pixel values of the working block. Implanting bits' number increases for a smaller value of range and vice versa. Being motivated, this research does the identical embedding task by the range of absolute-valued prediction errors. The research implements a predictor and measures the prediction errors in their absolute values. Computed errors' range is smaller than the range of the pixel values. The proposed method does the histogram-shifting task by the range of error values. The proposed method increases the number of embedded bits by a multiple of 1.31 to 2.04. In the second phase, it further applies the same predictor to the last calculated absolute-valued errors. The method yields a smaller range of absolute-valued errors by repeating the procedure for a number of times. The algorithm applies that smaller range in implementing HAM policy. The approach increases the number of implanted bits.