3D recognition and differentiation of objects

To identify, check or pick objects three-dimensionally in various applications, the correct recognition of the respective component is essential. Using the 3D scanner „Atos CORE” of the company GOM as well as the image processing library Halcon by MVTec, phil-vision developed a 3D matching system which is able to identify any object three-dimensionally.

For a start, every new part is placed under the scanner and recorded from different positions/perspectives. This way, the system generates from a 3D point cloud a reference model to be stored.

If another component is now placed under the scanner, it will again create a point cloud of the respective object. In the following, a 3D matching program constructed in Halcon checks the currently scanned component against all previously deposited models. If a model is found matching the object, the component will be recognized and nomenclatured with an initially defined naming. Depending on the desired matching score that is adjustable via a set of parameters, the software recognizes identical components with more or less similarities. Thus, slightly diverging objects can be classified as still acceptable or possibly sorted out. The information about the categorization of each part is put out to the customer’s control system, so it will be able to react accordingly.

The application possibilities of the system are various: Presence and completeness analysis, defect detection on surfaces, pick and place applications as well as the positioning of objects can, for example, be applications which precise 3D models are required for.