基于游程统计和Walsh谱能量分布的调色板图像隐写分析

摘要

提出一种基于游程统计和Walsh谱能量分布的调色板图像隐写分析方法,利用调色板颜色的奇偶值(parity value,P值)构造P值图像,通过游程统计和Walsh变换提取出46维特征,并用校准技术削弱载体图像差异对分析统计量的干扰.实验表明,该方法适用于EzStego、分量和以及最佳奇偶隐写等调色板图像隐写算法,比奇异颜色分析准确.%A new method is proposed for palette image steganalysis based on the run-length statistics and the Walsh spectrum energy distribution. The P-value binary image is constructed and then 46-dimensional features are extracted from the run-length histogram and Walsh spectrum coefficients. Calibration is utilized to eliminate varieties caused by hosts. Experiments show that our scheme can successfully detect EzStego, sum of RGB-based steganography, and OPA with low embedding rates and it is superior to some prior arts.