无失真图像OpenCV C++的SURF与匹配

Question

我正在研究OpenCV 4的ROS旋律。在undistort()之后，图像有一个由冲浪检测到的黑色背景。我怎么才能解决这个问题？

​

Answer 1

由于米卡的评论，我找到了解决办法。我在低比率测试中过滤了特征：

//-- Filter matches using the Lowe's ratio test
//Default ratio_thresh: 0.7f; 
vector<DMatch> matches;
size_t i = 0;
bool lowe_condition = false;
bool black_background_condition = false;

//Filter matches in black background
for (i; i < knn_matches.size(); i++)
{
    lowe_condition = (knn_matches[i][0].distance < ratio_thresh * knn_matches[i][1].distance);
    black_background_condition = ((keypoints1[i].pt.x >= width_low ) && (keypoints1[i].pt.x <= width_high)) && ((keypoints1[i].pt.y >= height_low ) && (keypoints1[i].pt.y <= height_high));

    if (lowe_condition && black_background_condition)
    {
        matches.push_back(knn_matches[i][0]);
    }
}