/*** ＠brief 归一化特征点到同一尺度（作为normalize DLT的输入）** [x' y' 1]' = T * [x y 1]' \n* 归一化后x', y'的均值为0，sum(abs(x_i'-0))=1，sum(abs((y_i'-0))=1** @param vKeys             特征点在图像上的坐标* @param vNormalizedPoints 特征点归一化后的坐标* @param T                 将特征点归一化的矩阵 左乘*/
void Initializer::Normalize(const vector<cv::KeyPoint> &vKeys, vector<cv::Point2f> &vNormalizedPoints, cv::Mat &T) {//第一步得到所有特征点的均值,并将所有点的均值为0float meanX = 0;float meanY = 0;for (int i = 0; i < vKeys.size(); i++) {meanX += vKeys[i].pt.x;meanY += vKeys[i].pt.y;}meanX /= vKeys.size();meanY /= vKeys.size();//第二步将所有点到原点的距离为根号2float meanDevX = 0;float meanDevY = 0;for (int i = 0; i < vKeys.size(); i++) {vNormalizedPoints[i].x = vKeys[i].pt.x - meanX;vNormalizedPoints[i].y = vKeys[i].pt.y - meanY;meanDevX += fabs(vNormalizedPoints[i].x); //fabs是求一个实数的绝对值  点到原点距离的累加meanDevY += fabs(vNormalizedPoints[i].y);}meanDevX /= vKeys.size(); //点到原点距离的平均值meanDevY /= vKeys.size();for (int i = 0; i < vKeys.size(); i++) {vNormalizedPoints[i].x /= meanDevX;vNormalizedPoints[i].y /= meanDevY;}//用于还原特征点到原始的坐标系,获得矩阵   |sX  0  -meanx*sX|  用于取逆     x    快速还原//                                  |0   sY -meany*sY|          *   y//                                  |0   0      1    |              1float sX = 1.0 / meanDevX;float sY = 1.0 / meanDevY;T = Mat::eye(3, 3, CV_32F);T.at<float>(0, 0) = sX;T.at<float>(0, 2) = -meanX * sX;T.at<float>(1, 1) = sY;T.at<float>(1, 2) = -meanY * sY;T.at<float>(2, 2) = 1;
}