One particular type of data that I find I increasingly deal with is 3D point cloud data. Point cloud data is fairly simple to grasp, it’s merely a “cloud” of points which each have their own individual coordinates in a Cartesian (think X, Y, Z) system. If that still doesn’t clear it up, watch this excellent video which shows off a cool example of point cloud data to model a shipping gallery. The sources of such point cloud datasets vary, including laser scanners, range cameras, or even your standard close-range photogrammetric techniques. In this post, I’m going to examine some typical techniques for fitting linear geometric shapes (lines, planes) to 3D point cloud data. It may get hairy with all the math involved, but I’ll try to keep the equations down where possible.