如何在matlab中找到具有n个公共元素的大行矩阵的行索引？

Question

我正在尝试查找索引具有 2 个公共元素的矩阵的行对。我有一个 3D 对象的三角剖分，想通过相邻三角形的角度过滤三角形。所以，我必须找到共享一条边的三角形。为此，我必须找到 .tri 文件中具有 2 个公共点的行。

我有 350 个左右的 .tri 文件，每个文件的亮度约为 7000x3。我找到了这个资源： MathWorks link 接受的答案需要我 mac 播放 15 分钟，而 Matt Figs 的答案每个 .tri 文件大约需要 8 分钟（我有 350！）。

我怎样才能完成这个过程？有人告诉我尝试用 C 编写这部分代码（我从未使用过 C）。我也考虑过尝试在 linux 服务器上安装 Matlab 并在那里运行安装它（我也从未这样做过）。关于如何用 C 编写程序或使用 AWS 服务器有什么建议吗？

3D 文件位于此处：3D files 这是我运行的代码：（任何清理它的建议也将不胜感激）

addpath('/Users/len/Desktop/javaplex/nonrigid3d')
files1 = dir('/Users/len/Desktop/javaplex/nonrigid3d/*.tri');    
files2 = dir('/Users/len/Desktop/javaplex/nonrigid3d/*.vert');

time=cputime;
for k =1:length(files1)    
    X=load(files1(k).name);
    Pt=load(files2(k).name);

    %define a sparse matrix for holding angles between 2-simplices
    W=sparse(length(X),length(X));

    % Find common vertices: try with C
    Xs = sort(X,2);
    S = size(X,1);
    L = S*(S+1)/2;
    F = sparse(L,L);
    cnt = 0;
    Ntoshare = 2;
    for ii = 1:S
        for jj = ii+1:S
            cnt = cnt + 1;
            if sum(ismembc(Xs(ii,:),Xs(jj,:)))==Ntoshare
                F(jj,ii) = 1;
            end
        end

    end

[I,J] = find(F); % pairs of rows that have 2 common vertices.
cmn=[I,J];

% Calculate angle and store in sparse matrix. row and col reference triangles
for i=1:length(cmn)
    normal1=cross(Pt(X(cmn(i,1),1),:)-Pt(X(cmn(i,1),2),:) ,Pt(X(cmn(i,1),1),:)-Pt(X(cmn(i,1),3),:));

    normal2=cross(Pt(X(cmn(i,2),1),:)-Pt(X(cmn(i,2),2),:), Pt(X(cmn(i,2),1),:)-Pt(X(cmn(i,2),3),:));

    theta = acos((dot(normal1,normal2)/(norm(normal1)*norm(normal2))));

    W(cmn(i,1),cmn(i,2))=theta;
end   

% save W as a txt file as nameofimage.txt in current directory
[i,j,val]=find(W);
data_dump=[i,j,val];
l=files1(k).name;
filename = strrep(l, '.tri','.txt');
dlmwrite(filename, data_dump, 'delimiter',' ');

Answer 1

已经有很多内置方法可以对三角形网格进行计算。看看 triangulation class。以下几行将像您的代码一样计算非边界三角形之间的角度。

%%// Generate triangulation data structure
Tri = triangulation(X, Pt);
%%// Find neighboring triangles
edgeAttachments = Tri.edgeAttachments(Tri.edges);
neighbors = cell2mat(edgeAttachments(cellfun(@numel, edgeAttachments)==2));
%%// Compute angles
normals1 = Tri.faceNormal(neighbors(:,1));
normals2 = Tri.faceNormal(neighbors(:,2));
angles = acos(dot(normals1, normals2, 2)); %// faceNormals are already normalized
%%// Generate the output you are looking for
data_dump = [neighbors, angles];

如何在matlab中找到具有n个公共元素的大行矩阵的行索引？

How to find row indices of large row matrix with n common elements in matlab?

c

matlab

matrix

time-complexity

amazon-web-services