資源簡介
1維的簡單LDA和2維LDA人臉識別的matlab代碼,注釋詳細,可以直接運行,非常好用,還有1維LDA算法講解,很適合入門的同學!

代碼片段和文件信息
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%a?simple?example?for?using?lDA?to?classification
%author?:huangsheng?@?iiec.cqu
%copy?right:iiec.cqu
%date:2010-4-6
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
clc;
clear;
%實驗數據來源http://people.revoledu.com/kardi/tutorial/LDA/Numerical%20Example.html
cls1_data=[2.95?6.63;?2.53?7.79;?3.57?5.65;3.16?5.47];%第一個類的訓練集
cls2_data=[2.58?4.46;?2.16?6.22;?3.27?3.52];%第二個類的訓練集
%求期望
E_cls1=mean(cls1_data);%第一類數據的期望矩陣
E_cls2=mean(cls2_data);%第二類數據的期望矩陣
E_all=mean([cls1_data;cls2_data]);%所有訓練集的期望矩陣
%%%%%%%%%%%%%%%%%%%%分類前畫圖%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for?i=1:4
?????plot(cls1_data(i1)cls1_data(i2)‘.r‘);
?????hold?on;
end;
plot(E_cls1(1)E_cls1(2)‘^r‘);
hold?on;
for?i=1:3
?????plot(cls2_data(i1)cls2_data(i2)‘*b‘);
?????hold?on;
end;
plot(E_cls2(1)E_cls2(2)‘^b‘);
plot(E_all(1)E_all(2)‘vc‘);
hold?on;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%計算類間離散度矩陣:
x1=E_all-E_cls1;
x2=E_all-E_cls2;
Sb=4*x1‘*x1/7+3*x2‘*x2/7;
%計算類內離散度矩陣
y1=0;
for?i=1:4
????y1=y1+(cls1_data(i:)-E_cls1)‘*(cls1_data(i:)-E_cls1);
end;
y2=0;
for?i=1:3
????y2=y2+(cls2_data(i:)-E_cls2)‘*(cls2_data(i:)-E_cls2);
end;
Sw=4*y1/7+3*y2/7;
%求最大特征值和特征向量
[VL]=eig(inv(Sw)*Sb);
[ab]=max(max(L));
newspace=V(:b);%最大特征值所對應的特征向量
new_cls1_data=cls1_data*newspace;%訓練后的數據集
new_cls2_data=cls2_data*newspace;%訓練后的數據集
%%%%%%%%%%%%%%%%%%畫圖代碼%%%%%%%%%%%%%%%%%
abort=V(:1);%
k=newspace(2)/newspace(1);
k2=abort(2)/abort(1);%較小特征值所對應的特征向量的斜率
b2=E_all(2)-k2*E_all(1);
plot([4?2][4*k2+b2?2*k2+b2]‘-b‘);%畫出較小特征值對應的特征向量
hold?on;
%plot(E_all(1)E_all(2)‘*r‘);%畫出總期望點
plot([06][06*k]‘-c‘);%畫出最大特征值對應的特征向量,即樣本所組成的線性空間所投影的子空間
axis([2?6?0?11]);%防止坐標系
hold?on;
%畫出樣本投影到子空間點
for?i=1:4
????temp=cls1_data(i:);
????newx=(temp(1)+k*temp(2))/(k*k+1);
????newy=k*newx;
????plot(newxnewy‘*r‘);
end;
for?i=1:3
????temp=cls2_data(i:);
????newx=(temp(1)+k*temp(2))/(k*k+1);
????newy=k*newx;
????plot(newxnewy‘ob‘);
end;
%預測
prediction=[2.81?5.46];
result=prediction*newspace;
temp=new_cls1_data-[result;result;result;result];
temp2=new_cls2_data-[result;result;result];
if(min(abs(temp))>min(abs(temp2)))
????output=‘該樣本屬于不合格產品‘
else
????output=‘該樣本屬于合格產品‘
end;
?屬性????????????大小?????日期????時間???名稱
-----------?---------??----------?-----??----
?????文件?????288362??2010-04-11?17:12??LDA\1DLDA\LDA.pdf
?????文件???????2480??2010-04-11?17:03??LDA\1DLDA\LDA_CODE_IIEC_CQU.m
?????文件????????492??2007-12-10?14:48??LDA\2DLDA\classif.m
?????文件???????1806??2013-04-03?18:09??LDA\2DLDA\main.m
?????文件????4121792??2008-03-04?18:10??LDA\2DLDA\orldata.mat
?????文件????????326??2008-01-03?16:07??LDA\2DLDA\sortem.m
?????文件????????619??2000-05-17?21:56??LDA\2DLDA\success.m
?????文件???????1508??2013-04-02?11:22??LDA\2DLDA\tdfda.m
?????目錄??????????0??2010-04-11?17:20??LDA\1DLDA
?????目錄??????????0??2013-04-02?11:23??LDA\2DLDA
?????目錄??????????0??2013-04-02?11:24??LDA
-----------?---------??----------?-----??----
??????????????4417385????????????????????11
- 上一篇:基于MATLAB的人臉識別
- 下一篇:基于MATLAB的語音共振峰的估計
評論
共有 條評論