我正在努力克服kinect v2的深度颜色。我正在使用录制的,而不是实时数据,因此我无法使用CoordinateMapper。 我正在采用这种方法:http://nicolas.burrus.name/index.php/Research/KinectCalibration 不幸的是我无法在任何地方找到kinect v2的失真参数,所以我正在跳过将原始深度值转换为米。任何人都可以提供吗?我的点云看起来很不错!但是我的色彩映射根本不起作用。为了调试,我尝试创建bgr图像的变换图像(color_img_transformed),但它看起来非常错误:
for (int y = 0; y < depth.rows; ++y)
{
for (int x = 0; x < depth.cols; ++x)
{
float depthValue = (float) depth.at<float>(y,x);
pcl::PointXYZRGBA pt;
if (depthValue == -1000.f || depthValue == 0.0f || boost::math::isnan(depthValue))
{
}
else
{
//depthValue *= 1000; // i think values are already in mm
pt.x = (static_cast<float>(x + tmp_x) - centerX) * scaleFactorX * depthValue;
pt.y = (centerY - static_cast<float>(y + tmp_y)) * scaleFactorY * depthValue;
pt.z = depthValue;
// new try to map - https://www.codefull.org/2016/03/align-depth-and-color-frames-depth-and-rgb-registration/
// Apply RGB intrinsics
Eigen::Vector4f tmp = extrinsics_ * Eigen::Vector4f(pt.x,pt.y,pt.z,1);
float x_ = (tmp[0] * fx_rgb / tmp[2]) + cx_rgb;
float y_ = (tmp[1] * fy_rgb / tmp[2]) + cy_rgb;
// "x" and "y" are indices into the RGB frame, but they may contain
// invalid values (which correspond to the parts of the scene not visible
// to the RGB camera.
// Do we have a valid index?
if (x_ > bgr.size().width || y_ > bgr.size().height || x_ < 1 || y < 1)
continue;
// Need some kind of interpolation. I just did it the lazy way
int round_x = (int) floor(x_);
int round_y = (int) floor(y_);
cv::Vec3b colorValue = bgr.at<cv::Vec3b>(round_y, round_x);
RGBValue color;
color.Red = colorValue[2];
color.Green = colorValue[1];
color.Blue = colorValue[0];
pt.rgba = color.long_value;
color_img_transformed.at<cv::Vec3b>(round_y, round_x)[0] = 255;//colorValue[0];
color_img_transformed.at<cv::Vec3b>(round_y, round_x)[1] = 0;//colorValue[1];
color_img_transformed.at<cv::Vec3b>(round_y, round_x)[2] = 0;
}
}
}
我不太确定我是否使用了正确的外在/内在函数。有人可以对他们发表评论吗?