matlab 神经网络 遥感图像分类 可用程序

clear all; %<----------------------------!运用BP网络进行图像分类---------------------------> %读入样本10,即遥感图像的背景 I=imread('10.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); %初始化输入矢量P和输出矢量T P=[]; T=[]; P=[R;G;B]; T=[0;0;0]; [m n]=size(P); T=concur(T,n); %读入样本图像0 I=imread('0.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.1;0.3;0.5]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像1 I=imread('1.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.2;0.4;0.6]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像2 I=imread('2.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.3;0.2;0.7]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像3 I=imread('3.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.4;0.4;0.7]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像4 I=imread('4.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.5;0.5;0.4]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像5 I=imread('5.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.6;0.9;0.2]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像6 I=imread('6.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.7;0.2;0.8]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像7 I=imread('7.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.8;0.3;0.5]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像8 I=imread('8.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[0.9;0.2;0.1]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %读入样本图像9 I=imread('9.jpg'); %将样本图像降维处理 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); P1=[R;G;B]; T1=[1;0.5;0.6]; P=[P,P1]; [m n]=size(P1); T1=concur(T1,n); T=[T,T1]; %创建一个前向神经网络 net=newff(minmax(P),[5,3],{'logsig','purelin'},'traingdx'); %设置训练参数 net.trainParam.show=50; net.trainParam.epochs=1000; %最大训练步数为1000 net.trainParam.goal=0.001; net=init(net); %对BP网络进行训练 net=train(net,P,T); for j=1:2 if j==1 var=input('是否对2000年崇明东滩遥感影像进行分类（是:输入1 ，否:输入2）'); elseif j==2 var=input('是否对2003年崇明东滩遥感影像进行分类（是:输入1 ，否:输入2）'); end %读入待分类遥感图像 if var==1 if j==1 I=imread('tm2000mask.jpg'); elseif j==2 I=imread('tm2003mask.jpg'); end %将彩色图像降维 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %将灰度值归一化处理 R=im2double(R); G=im2double(G); B=im2double(B); [M,N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); p=[R;G;B]; %对BP网络进行仿真 Y=sim(net,p); R=Y(1,:); X=R; classR=[]; for i=0:(M-1) classR=[classR;R((i*N+1):(i*N+N))]; end G=Y(2,:); classG=[]; for i=0:(M-1) classG=[classG;G((i*N+1):(i*N+N))]; end B=Y(3,:); classB=[]; for i=0:(M-1) classB=[classB;B((i*N+1):(i*N+N))]; end R=abs(classR)*255; R=uint8(R); G=abs(classG)*255; G=uint8(G); B=abs(classB)*255; B=uint8(B); classify=cat(3,R,G,B); figure,imshow(classify); if j==1 title('2000年崇明东滩遥感图像分类图'); elseif j==2 title('2003年崇明东滩遥感图像分类图'); end %计算各类地物面积 X=abs(X)*255; [M,N]=size(X); t10=0;t0=0;t1=0;t2=0;t3=0;t4=0;t5=0;t6=0;t7=0;t8=0;t9=0; for i=1:N if X(1,i)==0 t10=t10+1; elseif X(1,i)<26 t0=t0+1; elseif X(1,i)<52 t1=t1+1; elseif X(1,i)<=76 t2=t2+1; elseif X(1,i)<=102 t3=t3+1; elseif X(1,i)<=127 t4=t4+1; elseif X(1,i)<=153 t5=t5+1; elseif X(1,i)<=178 t6=t6+1; elseif X(1,i)<204 t7=t7+1; elseif X(1,i)<=229 t8=t8+1; else t9=t9+1; end end var=input('是否以饼状图显示地物比例(是:输入1 ,否:输入2)'); if var==1 %绘制地物面积饼状图 t=[t0,t1,t2,t3,t4,t5,t6,t7,t8,t9]; figure; pie(t); legend('海水','农地','绿林地','房屋','养殖场','芦苇','互花米草','海三棱藨草','光滩','未利用地',-1); %标注图例 end var=input('是否以直方图显示地物比例(是:输入1 ,否:输入2 )'); if var==1 %绘制地物面积柱状图 y=[t0,t1,t2,t3,t4,t5,t6,t7,t8,t9]; x=1:10; figure; bar(x,y);colormap summer; %绘制柱状图且返回图形句柄 end end end var=input('是否对2005年遥感图像进行预测(是：输入1 否：输入2)'); if var==1 %<----------------------------!运用2000年和2003年图像预测2005年图像------------------> %读入2000年遥感数据 I=imread('tm2000mask.jpg'); %将图像数据降维 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); %初始化输入向量 P=[]; P=[R;G;B]; %初始化输出向量 %读入2003年遥感数据 I=imread('tm2003mask.jpg'); %将图像数据降维 R=I(:,:,1); G=I(:,:,2); B=I(:,:,3); %灰度值归一化 R=im2double(R); G=im2double(G); B=im2double(B); [M N]=size(R); R=reshape(R',[1 M*N]); G=reshape(G',[1 M*N]); B=reshape(B',[1 M*N]); T=[]; T=[R;G;B]; %创建一个前向神经网络 net=newff(minmax(P),[5 3],{'logsig','purelin'},'traingdx'); %设置训练参数 net.trainParam.show=50; net.trainParam.epochs=1000; net.trainParam.goal=0.001; %对BP网络进行训练 net=train(net,P,T); %预测2005年遥感图像 %对BP网络进行仿真 Y=sim(net,T); R=Y(1,:); R2005=[]; for i=0:(M-1) R2005=[R2005;R((i*N+1):(i*N+N))]; end G=Y(2,:); G2005=[]; for i=0:(M-1) G2005=[G2005;G((i*N+1):(i*N+N))]; end B=Y(3,:); B2005=[]; for i=0:(M-1) B2005=[B2005;B((i*N+1):(i*N+N))]; end R=abs(R2005)*255; R=uint8(R); G=abs(G2005)*255; G=uint8(G); B=abs(B2005)*255; B=uint8(B); T2005=cat(3,R,G,