Lossless Authentication Watermarking Based on Adaptive Modular Arithmetic

Abstract

Reversible watermarking schemes based on modulo-256 addition may cause annoying salt-and-pepper noise. To avoid the salt-and-pepper noise, a reversible watermarking scheme using human visual perception characteristics and adaptive modular arithmetic is proposed. First, a high-bit residual image is obtained by extracting the most significant bits (MSB) of the original image, and a new spatial visual perception model is built according to the high-bit residual image features. Second, the watermark strength and the adaptive divisor of modulo operation for each pixel are determined by the visual perception model. Finally, the watermark is embedded into different least significant bits (LSB) of original image with adaptive modulo addition. The original image can be losslessly recovered if the stego-image has not been altered. Extensive experiments show that the proposed algorithm eliminates the salt-and-pepper noise effectively, and the visual quality of the stego-image with the proposed algorithm has been dramatically improved over some existing reversible watermarking algorithms. Especially, the stegoimage of this algorithm has about 9.9864 dB higher PSNR value than that of modulo-256 addition based reversible watermarking scheme.