如何从深度相机获取世界系统中的深度坐标

Question

depth camera: intel realsense D415 language: python

I am trying to get the z coordinate of a depth point in the world coordinate system. I am wondering if there's an embedded method or if there is a way to obtain that?

Thank you in advance!

Answer 1

It sounds like you want to use a separate and fixed object in the scene as a reference, eg a table supporting your objects of interest.

Finding the 3D plane of the table is usually easy: you use some robust fitting algorithm along with reasonable priors (eg the largest plane, or the largest plane approximately oriented in a known way with respect to the camera).

Finding a plane, however, only gives you a "z" direction orthogonal to it, plus a translation vector to a point on the plane which may not be a desirable one. So you need a way to identify some desirable "origin" point in the point cloud. This is usually done using an object (aka "calibration jig" or "rig") of known shape that can easily be identified and precisely fit to a model. For example, a billiard ball (fit a sphere, find the center, offset by the radius to the contact point with the plane), or a cone (find the center of the base), etc. Note that using a conical jig allows to find both plane and point in one shot.