基于matlab实现的sfs算法对目标的多角度图像进行三维重构.rar

tic %%%%%%%%每张图中光源的方向 light1=[0 -10 0]; light2=[-10 -10 0]; light3=[10 -10 0]; light4=[0 -10 10]; light5=[-10 -10 10]; light6=[10 -10 10]; %测试结束 %对光源方向进行归一化 light1 = light1 / norm(light1); light2 = light2 / norm(light2); light3 = light3 / norm(light3); light4 = light4 / norm(light4); light5 = light5 / norm(light5); light6 = light6 / norm(light6); %%将六张二维图读进来，注意图像必须是灰度图 img1 = rgb2gray( imread('vase1.png')); img2 = rgb2gray( imread('vase2.png')); img3 = rgb2gray( imread('vase3.png')); img4 = rgb2gray( imread('vase4.png')); img5 = rgb2gray( imread('vase5.png')); img6 = rgb2gray( imread('vase6.png')); %设置光源矩阵 S= [light1; light2 ;light3; light4; light5; light6]; %初始化b,b存储每一个像素点分别沿x,y,z轴的反射率 b=ones(800,800,3); b=double(b); %初始化p,q,p,q分别存储沿x轴，y轴的单位反射率，也就是沿x,y的梯度 p=ones(800,800); p=double(p); q=p; %初始化Z,Z存储图像中每一个像素点的高度值 Z=ones(800,800); Z=double(Z); for i=1:800 for j=1:800 E=[img1(i,j) img2(i,j) img3(i,j) img4(i,j) img5(i,j) img6(i,j)]; E=double(E'); tb= (inv(S'*S))*S'*E; %tb是3*1矩阵，求的是目标本身的反射率 nbm = norm(tb);%求tb的范数 if( nbm == 0) b(i,j,:) = 0; else b(i,j,:) = tb / nbm; %b存储归一化后的反射率 end %tM是个中间变量，存储一个点沿x,y,z轴的反射率 tM = [b(i,j,1) b(i,j,2) b(i,j,3)]; nbm = norm(tM); if( nbm == 0) tM = [0 0 0]; else tM = tM / nbm; end p(i,j)=tM(1,1);%p是改点沿x轴的单位反射率，也就是沿x轴的梯度 q(i,j)=tM(1,2);%q是改点沿y轴的单位反射率，也就是沿y轴的梯度 end end %对p,q进行积分，就可求出每一个像素点的高度值 for i=1:800 for j=1:800 Z(i,j) = sum(q(1:i, 1)) + sum(p(i,1:j)); end end Z = Z*-1; figure(1); hold on; for i=1:2:800 for j=1:2:800 plot3(j+b(i,j,1),i+b(i,j,2),b(i,j,3),'b' );%显示梯度图像 end end hold off; figure(2); mesh(Z); toc