Visualizing 3D point clouds with PCL

While working with 3D data, it is hard to quickly understand if a result is correct simply by looking at reprojection error measures or raw point information. On the other hand, if we look at the point cloud itself we can immediately verify whether it makes sense or there was an error. For visualization we will use an up-and-coming sister project for OpenCV, called the Point Cloud Library (PCL). It comes with many tools for visualizing and also analyzing point clouds, such as finding flat surfaces, matching point clouds, segmenting objects, and eliminating outliers. These tools are highly useful if our goal is not a point cloud but rather some higher-order information such as a 3D model.

First, we should represent our cloud (essentially a list of 3D points) in PCL's data structures. This can be done as follows:

pcl::PointCloud<pcl::PointXYZRGB>::Ptr cloud;

void PopulatePCLPointCloud(const vector<Point3d>& pointcloud,
const std::vector&lt...