Best way to get good features

edit

Hi all. I'm extracting features from images. On the Opencv example, good matches are extracted in this way:

for (int i = 0; i < descriptors_object.rows; i++)
{
    if (matches[i].distance <= 3 * min_dist)
    {
        good_matches.push_back(matches[i]);
    }
}

(I'm taking matches array with knn)

but it doesn't take alway good matches. Because I need a precise method, do you know a better way to get good matches?