Abstract: The goal of the steganalysis is detecting and making the hidden information useless. For its significance in national security, copyright protection et., it attracts more and more researchers interests. This paper proposes a feature based blind steganalysis to determine the existence of the hidden information, and locate the hidden domain. It focuses on the QIM based methods with which secret message is embedded, and the statistical features are obtained from the DCT and wavelet domain. For accurate classification, it adopts neural network, and uses an image database composed of 360 plain and stego-images for evaluating. The experimental results based on this database show that the proposed method is valid.

Key words: blind steganalysis, Quantization Index Modulation (QIM), histogram analysis

摘要： 提出一种基于频谱分析和神经网络分类的盲检测算法，该算法不仅较准确地判断出图像中是否存在隐藏信息，而且可以确定信息隐藏的位置。针对基于QIM方法的信息隐藏，检测时所使用的特征量分别从DCT域和小波域中提取。使用标准图像集所做的实验结果验证了该算法的有效性。

关键词: 盲检测, 量子索引调制, 直方图分析

CLC Number: 

• TP309