摘要： 在图像隐秘检测中，图像特征形成是关键技术之一。该文建立了自然图像的小波包分解系数绝对值线性预测模型，提出了基于小波包变换的图像特征形成算法。为解决图像特征向量维数随小波包分解层数呈指数级增长而导致的“维数灾难”问题，给出了基于离散度的可分性准则的详细计算过程。

关键词: 隐秘检测, 特征形成, 小波包变换, 线性预测模型

Abstract: The creation of images characteristics is one of the key technologies to the detection of image steganography. In this article, the linear magnitude predictor of wavelet packet decomposition coefficients, the creation algorithm of images characteristics based on wavelet packet transform have been introduced. To solve the problem of dimension disaster caused by the exponential increase of dimensions of image characteristics vector with the increase of the level of wavelet packet decomposition, the detail of calculation of separability rule based on scatter has been introduced.

中图分类号: 

• TP391.41