基于散度分析的模拟图像信号特征选择  

Based on Analysis of the Divergence of Analog Image Signal Feature Selection

作　　者：张军丽[1] 

机构地区：[1]郑州大学西亚斯国际学院,河南郑州451150

出　　处：《计算机仿真》2015年第1期434-437,共4页Computer Simulation

摘　　要：在对模拟图像信号进行特征选择的过程中,容易出现各类别与模拟图像信号特征之间的关系考察不充分的情况,导致传统的基于优化蚁群算法的模拟图像信号特征选择方法,由于只在整个类标签集的角度上进行特征选择,无法有效实现模拟图像信号特征的选择,提出一种基于散度分析的模拟图像信号选择方法,建立特征选择问题的数学模型,依据Fisher准则的分类思想,将散度差作为模拟图像信号特征选择的准则,获取一种基于散度差的模拟图像信号特征分类选择方法,将加权矩阵散度差的迹看作是投影鉴别规则,通过获取一组最优投影方向,使得各种类别的原始向量经投影后可最大程度的被分离,对模拟图像信号特征值对应的特征向量进行标准正交化操作,从而实现模拟图像信号的特征选择。仿真实验结果表明,所提方法具有很高的准确性。In the simulation image signal in the process of feature selection, don＇t and prone to all kinds of modeling of the relationship between the characteristics of image signal, insufficient in traditional simulation based on optimization ant colony algorithm, the image signal feature selection methods, because only in the Angle of the whole class labels on feature selection, cannot effectively choose analog image signal characteristics, puts forward a simulation image signal selection method based on the divergence analysis, establish the mathematics model of feature selection problems, according to the classification of the Fisher criterion, the divergence difference as the simulation image signal feature selection criterion, a divergence difference based on the simulation of image signal feature classification method, the weighted matrix differential divergence trace as projective identification rules, by getting a group of opti- mal projection direction, making various kinds of original vector after the projection can be separated to a great extent, to the analog image signal eigenvalue corresponding eigenvector for orthonormalization operation, so as to realize the feature selection of simulated image signal. The simulation results show that the proposed method has high accuracy.

关 键 词：散度分析 模拟图像信号 特征选择 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]