bayes.zip

%% 第十二章 对象识别 %% 例12.1 gscale dftcorr 使用相关来匹配图像 clc clear f = imread('..\Pictures\images_ch12\Fig1201(a)(Hurricane Andrew).tif'); imshowMy(f) title('原始图像') w = imread('..\Pictures\images_ch12\Fig1201(b)(hurricane_mask).tif'); imshowMy(w) title('模板图像') g = dftcorr(f,w); gs = gscale(g); imshowMy(gs) [I, J] = find(g == max(g(:))) imshowMy(gs > 254) %% 例12.2 多谱数据的贝叶斯分类 clc clear f1 = imread('..\Pictures\images_ch11\Fig1125(a)(WashingtonDC_Band1_512).tif'); f2 = imread('..\Pictures\images_ch11\Fig1125(b)(WashingtonDC_Band2_512).tif'); f3 = imread('..\Pictures\images_ch11\Fig1125(c)(WashingtonDC_Band3_512).tif'); f4 = imread('..\Pictures\images_ch11\Fig1125(d)(WashingtonDC_Band4_512).tif'); B1 = roipoly(f1); B2 = roipoly(f1); B3 = roipoly(f1); stack = cat(3,f1,f2,f3,f4); [X1,R1] = imstack2vectors(stack, B1); [X2,R2] = imstack2vectors(stack, B2); [X3,R3] = imstack2vectors(stack, B3); Y1 = X1(1:2:end,:); Y2 = X2(1:2:end,:); Y3 = X3(1:2:end,:); [C1,m1] = covmatrix(Y1); [C2,m2] = covmatrix(Y2); [C3,m3] = covmatrix(Y3); CA = cat(3,C1,C2,C3); MA = cat(2,m1,m2,m3); dY{1} = bayesgauss(Y1,CA,MA); dY{2} = bayesgauss(Y2,CA,MA); dY{3} = bayesgauss(Y3,CA,MA); IY{1} = find(dY{1} ~=1 ); IY{2} = find(dY{2} ~=2 ); IY{3} = find(dY{3} ~=3 ); %------------------------------------ 训练模式 QQ_training = zeros(3,5); for i = 1:3 QQ_training(i,4) = length(dY{i}); for j = 1:3 QQ_training(i,j) = length(find(dY{i} == j )); end QQ_training(i,5) = 100*(1 - length(IY{i})/length(dY{i})); end QQ_training %------------------------------------ 独立模式 Y1 = X1(2:2:end,:); Y2 = X2(2:2:end,:); Y3 = X3(2:2:end,:); dY{1} = bayesgauss(Y1,CA,MA); dY{2} = bayesgauss(Y2,CA,MA); dY{3} = bayesgauss(Y3,CA,MA); IY{1} = find(dY{1} ~=1 ); IY{2} = find(dY{2} ~=2 ); IY{3} = find(dY{3} ~=3 ); QQ_test = zeros(3,5); for i = 1:3 QQ_test(i,4) = length(dY{i}); % 样本总数 for j = 1:3 QQ_test(i,j) = length(find(dY{i} == j )); end QQ_test(i,5) = 100*(1 - length(IY{i})/length(dY{i})); % 正确率 end QQ_test %% 例12.2 多谱数据的贝叶斯分类(显示了错误点) clc clear f1 = imread('..\Pictures\images_ch11\Fig1125(a)(WashingtonDC_Band1_512).tif'); f2 = imread('..\Pictures\images_ch11\Fig1125(b)(WashingtonDC_Band2_512).tif'); f3 = imread('..\Pictures\images_ch11\Fig1125(c)(WashingtonDC_Band3_512).tif'); f4 = imread('..\Pictures\images_ch11\Fig1125(d)(WashingtonDC_Band4_512).tif'); B1 = roipoly(f1); B2 = roipoly(f1); B3 = roipoly(f1); B = B1|B2|B3; imshowMy(B) stack = cat(3,f1,f2,f3,f4); [X1,R1] = imstack2vectors(stack, B1); [X2,R2] = imstack2vectors(stack, B2); [X3,R3] = imstack2vectors(stack, B3); Y1 = X1(1:2:end,:); Y2 = X2(1:2:end,:); Y3 = X3(1:2:end,:); [C1,m1] = covmatrix(Y1); [C2,m2] = covmatrix(Y2); [C3,m3] = covmatrix(Y3); CA = cat(3,C1,C2,C3); MA = cat(2,m1,m2,m3); dY{1} = bayesgauss(Y1,CA,MA); dY{2} = bayesgauss(Y2,CA,MA); dY{3} = bayesgauss(Y3,CA,MA); IY{1} = find(dY{1} ~=1 ); IY{2} = find(dY{2} ~=2 ); IY{3} = find(dY{3} ~=3 ); %------------------------------------ 训练模式 QQ_training = zeros(3,5); for i = 1:3 QQ_training(i,4) = length(dY{i}); for j = 1:3 QQ_training(i,j) = length(find(dY{i} == j )); end QQ_training(i,5) = 100*(1 - length(IY{i})/length(dY{i})); end % %------------------------------------ 独立模式 % Y1 = X1(2:2:end,:); % Y2 = X2(2:2:end,:); % Y3 = X3(2:2:end,:); % % dY{1} = bayesgauss(Y1,CA,MA); % dY{2} = bayesgauss(Y2,CA,MA); % dY{3} = bayesgauss(Y3,CA,MA); % % IY{1} = find(dY{1} ~=1 ); % IY{2} = find(dY{2} ~=2 ); % IY{3} = find(dY{3} ~=3 ); % % QQ_test = zeros(3,5); % % for i = 1:3 % QQ_test(i,4) = length(dY{i}); % for j = 1:3 % QQ_test(i,j) = length(find(dY{i} == j )); % end % QQ_test(i,5) = 100*(1 - length(IY{i})/length(dY{i})); % end www1 = R1(IY{1},:); for i=1:length(www1) B(www1(i,1),www1(i,2)) = 0; end www2 = R2(IY{2},:); for i=1:length(www2) B(www2(i,1),www2(i,2)) = 0; end www3 = R3(IY{3},:); for i=1:length(www3) B(www3(i,1),www3(i,2)) = 0; end imshowMy(B) %% 例12.3 基于符号串匹配的对象识别（目前无法运行） clc clear A = imread('..\Pictures\images_ch12\Fig1203(a)(bottle_1).tif'); imshowMy(A) B = imread('..\Pictures\images_ch12\Fig1203(d)(bottle_2).tif'); imshowMy(B) b = boundaries(B,4,'cw'); b = b{1}; [M,N] = size(B); xmin = min(b(:,1)); ymin = min(b(:,2)); [xA,yA] = minperpoly(A,2); b2 = connectpoly(xA,yA); B2 = bound2im(b2,M,N,xmin,ymin); imshowMy(B2) [xB,yB] = minperpoly(B,2); [xn,yn] = randvertex(x,y,npix); angles = polyangles(x,y); s = floor(angles/45) + 1; s = int2str(s); R = strsimilarity(s11,s12); %% clc clear %% clc clear