Abstract:
This paper demonstrates and analyzes the affect on steganalysis detection precision rate during image steganalysis detection process caused by feature dimensions, and contrasts the recognition rate of artificial selection with dimensionality reduction algorithm. Result shows that lower dimensionalities of the image feature vector not only make the classifier more effective and reduce the computational complexity, but also improve the image steganalysis detection rate, and the manifold learning algorithm is better than artificial selection in dimensionality reduction.
Key words:
steganalysis detection,
feature extraction,
dimensionality reduction
摘要: 通过实验验证并分析图像隐写检测过程中特征维数对隐写检测正确率的影响,对比使用人工选取与机器降维的隐写图像识别率。结果表明,低维特征更有利于简化分类器的设计,降低计算复杂度,提高隐写检测正确率,且机器降维后的特征相比人工选取的特征拥有更好的隐写检测效果。
关键词:
隐写检测,
特征选取,
降维
CLC Number:
XIANG Li, BO Feng, SU Guang-Wei, SHEN Jun-Wei. Influence Analysis of Feature Dimension on Steganalysis Detection[J]. Computer Engineering, 2010, 36(21): 132-133,136.
相丽, 潘峰, 苏光伟, 申军伟. 特征维数对隐写检测的影响分析[J]. 计算机工程, 2010, 36(21): 132-133,136.