Abstract: Applying Tian's difference expansion to color images, this paper proposed a lossless data hiding method for color images based on prediction error difference expansion. To solve the quality decline of stego-images from excessive modulation to pixels, the proposed method dispersed smaller expansion to two color components, and improved histogram shifting technique. Experimental results show that both the capacity for embedding users' message and the stego-image quality are significantly improved, when compared with other algorithms.

Key words: reversible data hiding, prediction error difference expansion, Peak Signalto Noise Ratio (PSNR), color component

摘要： 将Tian差值扩展技术应用于彩色图像中，提出一种利用预测误差差值进行扩展嵌入的彩色图像无损数据隐藏算法。传统的差值扩展和预测误差扩展嵌入技术最大的缺点是过分利用差值造成载体图像质量严重下降。针对这一问题，该方法利用色彩分量间的相关性减小差值，并将差值扩展量分散到两个色彩分量中，同时，对差值直方图平移技术进行改进，使得同等嵌入率下图像质量达到最佳。实验结果表明，与其他算法相比，本算法在嵌入率和图像质量方面都有明显提高。

关键词: 无损数据隐藏, 预测误差差值扩展, 峰值信噪比, 色彩分量