LDA算法的matlab源代码

使用LDA（线性判别分析）算法提取一维数字信号（数组）的特征，可用于信号的分类识别。

function [V_selectFV1D1] = TYZmyLDA（MatdogMathum）
[rowdcold]=size（Matdog）;
[rowhcolh]=size（Mathum）;
n=rowd+rowh;
MeanDog=mean（Matdog）;
MeanHum=mean（Mathum）;
Xdog=Matdog-repmat（MeanDogrowd1）;
Xhum=Mathum-repmat（MeanDogrowh1）;
%狗的类内散度矩阵
for i=1:rowd
SwDog=+（Xdog（i:））‘*Xdog（i:）;
end
%人的类内散度矩阵
for j=1:rowh
SwHum=+（Xhum（j:））‘*Xhum（j:）;
end
Sw=（SwDog+SwHum）;%类内散布矩阵
%下面计算类间散布矩阵Sb
Mat=[Matdog;Mathum];
MeanMat=mean（Mat）;
D11=MeanDog-MeanMat;
D22=MeanHum-MeanMat;
Sb=（rowd*D11‘*D11+rowh*D22‘*D22）;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%相位使用1025
Sbw=Sb-Sw+eye（1025）;%%%%%%%%%%避免矩阵奇异，出现复数特征值
[V1 D1] = eigs（Sbw）;
V_select=abs（V1（:4:6））;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%频谱使用411-5HZ的特征
% Sbw=Sb-Sw+eye（41）;%%%%%%%%%%避免矩阵奇异，出现复数特征值
% [V1 D1] = eigs（Sbw25）;
% V_select=abs（V1）;
%%%%%%%%%%%%%%%%%%
V=V1;
D=D1;
F=Mat*V_select;
%%%%%%%%%%%%%%%%%%%%%%%----end----%%%%%%%%%%%%%%%%%%%%%%%%%%%

 属性            大小     日期    时间   名称
----------- ---------  ---------- -----  ----

     文件       1047  2013-07-10 17:28  TYZmyLDA.m

----------- ---------  ---------- -----  ----

                 1047                    1