如何使用模板函数从缓冲区(T*数据数组)创建cv :: Mat？

Question

我想编写一个函数模板,并将其复制指针引用数据T* image来cv::Mat.我很困惑如何推广T和cv_type匹配.

template<typename T>
cv::Mat convert_mat(T *image, int rows, int cols) {
    // Here we need to match T to cv_types like CV_32F, CV_8U and etc.
    // The key point is how to connect these two
    cv::Mat mat(rows, cols, cv_types, image);
    return mat;
}

我是模板编程的新手,我很困惑如何实现T-cv_types对应.

任何人有任何想法？谢谢!!!

Answer 1

使用cv::DataType<T>::type.

这是一个例子.

// Create Mat from buffer 
#include <opencv2/opencv.hpp>
using namespace std;
using namespace cv;


/*
//! First version
//! 2018.01.11 21:16:32 (+0800)
template <typename T>
Mat createMat(T* data, int rows, int cols) {
    // Create Mat from buffer
    Mat mat(rows, cols, cv::DataType<T>::type);
    memcpy(mat.data, data, rows*cols * sizeof(T));
    return mat;
}
*/

//! Second version 
//! 2018.09.03 16:00:01 (+0800) 
template <typename T>
cv::Mat createMat(T* data, int rows, int cols, int chs = 1) {
    // Create Mat from buffer 
    cv::Mat mat(rows, cols, CV_MAKETYPE(cv::DataType<T>::type, chs));
    memcpy(mat.data, data, rows*cols*chs * sizeof(T));
    return mat;
}

int main(){
    int    arr1[4] = {1,2,3,4};
    double arr2[4] = {1.1,2.2,3.3,4.4};

    Mat mat1 = createMat<int>(arr1, 2,2);
    Mat mat2 = createMat<double>(arr2, 2,2);
    cout << "Mat1:\n"<< mat1 <<endl;
    cout << "Mat2:\n"<< mat2 <<endl;
}

结果:

Mat1:
[1, 2;
 3, 4]
Mat2:
[1.1, 2.2;
 3.3, 4.4]