Question

opencv pca输出的特征值都为0

我想将5*3的矩阵降维成5*1矩阵，调试时发现mean的3个值都是0，请问这是为什么呢？
int n = 5;
int oldDim = 3;
int newDim = 1;
void main()
{

int i,j;
ifstream inTrainMat("D:/myProjects/ORB/ORB/result/pca/try.txt");
Mat trainMat = Mat::Mat( n,oldDim,CV_32FC1);
for (i=0;i<n;i++)
{
for (j=0;j<oldDim;j++)
{
int tmp;
inTrainMat>>tmp;
trainMat.at<int>(i,j) = tmp;
}
}

PCA pca(trainMat, Mat(), CV_PCA_DATA_AS_ROW, newDim);
Mat mean = pca.mean.clone();
Mat eigenvalues = pca.eigenvalues.clone();
Mat eigenvectors = pca.eigenvectors.clone();
for (i=0;i<oldDim;i++)
{
cout<<eigenvalues.at<float>(i,0)<<endl;
}
}
文件try.txt中的数据是：
23 3 9
8 1 77
1 4 6
4 28 5
6 7 3

Answer 1

这句
trainMat.at<int>(i,j) = tmp;
改为
trainMat.at<float>(i,j) = tmp;