3D in 2D Images – Part IV – Calculating the 3D pose of any 2D polygon

Posted on March 24, 2012 by Thor

With the techniques shown in “3D in 2D Images – Part III” we have the ability to create the 3D pose of any object represented by a 2D polygon – we can call it PolygonMatch3D based on the very powerful Scorpion Vision Software tool PolygonMatch which locates any occluded polygon with sub-pixel accuracy in 2D

In “3D in 2D Images – Part III” we show that we can locate a circle based on 4-6 user selectable edge-points – see image 1.

Image 1 - By locating the four edge-points in 3D - i have the accurate 3D plane of the cylinder 🙂


The four to six edge points, yellow markers, can be placed on any shaped object represented by a 2D polygon to calculate it’s 3D Pose. In Image 2 the cylinder’s 3D pose is visualised based on the fast and accurate 3D pose estimation methods developed based on  Scorpion Vision 9.2.

Image 2 - The 3D pose of the cylinder is estimated

What can we do?

We can locate the pose of any 2D polygon in 3D.

  1. We estimate the 3D Pose of a Circle or an ellipse
  2. We can estimate the 3D Pose of an object located by a Blob
  3. We can locate of any object that we can represent by a 2D polygon

Filed under: Machine Vision, Scorpion 9, Scorpion Vision 3D, Scorpion Vision Software, Smart Camera | Tagged: |

« »
%d bloggers like this: