Reversible Watermarking dengan Metode Modifikasi Histogram pada Difference Image

Abstract

Abstract— Reversible watermarking schemes are widely used to maintain the authenticity of the digital image. This research will discuss a method on histogram modification of difference image in which the difference image is created from the difference value of adjacent pixels of the image grayscale. 

Embedding process begins by dividing the host image and watermark into b blocks, followed by making a difference image of the host image block. From the difference image histogram, determine the peak value, and modify the histogram based on the peak value. Then, insert each block of the watermark to the difference image that has been modified and transform back into the grayscale image.

Extraction and recovery process is the inverted version of the embedding stage. This process begins by dividing the watermarked image into b blocks, followed by making a difference image of each block. Then, extract the data and shift difference image histogram using a peak value. That difference image is transformed back into a grayscale image.

Experimental results demonstrate that the average insertion capacity is 14% greater than Xue's with PSNR value over 48 dB for 4 x 4 pixels and 23 % greater with PSNR over 46 dB for 8 x 8 pixels. From the comparison of robustness to line and salt n’ pepper on the density of 0.05 noises is obtained that the watermark with ECC is more robust than a watermark without ECC. JPEG compression in lossless mode may be applied to the watermarked image. Multiple insertion of watermark can be done with the consequence that the more insertions will result in lower PSNR values.

Keywords—difference image, histogram modification, reversible watermarking