One of the most exciting features introduced in HALCON 10 was the ability to locate objects in a 3D point cloud based on matching their 3D surface profiles to a trained 3D model. While this technique is excellent for finding rigid parts that don’t deviate from the model, it can’t reliably find deformed objects or those that may have naturally occurring variations in shape or size.
In HALCON 12, surface-based 3D matching has been extended with a new technique called deformable surface-based matching which can locate objects with moderate deformations. Stretched or bent objects such as bananas, pretzels, or pliable materials can now be reliably located in a 3D scene despite moderate deformations, partial occlusions, or being located in a cluttered and/or noisy scene.
For information on how the deformable matching technique works and how to apply it in an application, read section 4.4 of the 3D Vision Solution Guide
