Using a C ++ interface might be more appropriate. Note that in the code example in the documentation [1] there is no modulo operation and, therefore, is not executed.
const int dims = 2;
int size[] = {3, 20};
SparseMat sparse_mat(dims, size, CV_32F);
for(int i = 0; i < 1000; i++) {
int idx[dims];
for(int k = 0; k < dims; k++)
idx[k] = rand() % size[k];
sparse_mat.ref<float>(idx) += 1.f;
}
cout << "bottom right element @ (2,19) = " << sparse_mat.ref<float>(2,19) << "\n";
Mat dense;
sparse_mat.convertTo(dense, CV_32F);
cout << dense;
Gives the next exit
bottom right element @ (2,19) = 19
[9, 23, 13, 26, 18, 13, 18, 15, 13, 17, 13, 18, 19, 6, 20, 20, 12, 15, 15, 15;
17, 17, 14, 16, 12, 14, 17, 15, 15, 18, 24, 18, 13, 22, 18, 11, 18, 22, 17, 15;
19, 16, 14, 10, 18, 19, 10, 17, 18, 15, 24, 22, 18, 18, 18, 23, 21, 16, 14, 19]
[1] Reference guide to OpenCV. Version 2.4.3. 2012, p. 46.