3D image processing

The new generation of 3D sensors has brought about a significant increase in the use of 3D image processing systems in manufacturing. Fast acquisition times of dense measuring points below one second are now easily achieved with the 3D sensors on today’s market. This opens up new fields of application, both in 3D measurement and testing technology for quality assurance as well as for 3D object recognition for implementing automated solutions. Besides appropriate sensor systems, fast and intelligent 3D data analysis techniques are needed in order to implement industrial applications. We offer a wide range of algorithms for this. This includes various fit algorithms and automated segmentation methods, which are particularly suitable for point cloud processing. Calibration procedures that enable the versatile use of new sensor technologies are also available, e.g. time-of-flight sensor systems or 3D smart cameras.
We are increasingly using machine learning and artificial intelligence in new developments for 3D image and data analysis.

A core area of 3D image processing is object recognition:
3D object recognition is a key success factor when it comes to implementing automation solutions for handling parts or creating a digital twin of a manufacturing environment. In addition to appropriate sensors for optical data acquisition, fast and intelligent object recognition methods are required above all to implement robust industrial applications. In this field, we offer a wide range of classical and AI-based methods.

 

3D fit methods

For many tasks in industrial image processing and coordinate measuring technology, fitting curves and surfaces into measuring points plays a major role. Geometric fitting is also an important sub-step for object recognition and scene analysis.

 

3D object recognition

To automate production processes, it is essential to recognize objects and their position. This calls for a precise 3D object recognition technology that can react flexibly to changes in the process and be implemented efficiently using robots.

 

Virtual assembly

When assembling several components, the deviations of the individual components accumulate, which can result in non-functional products. The goal of virtual assembly is the complete use of the recorded geometric measurement data of the components and the simulation of the assembly.

 

Segmentation methods

Optical sensor systems, such as laser line scanners or fringe projection systems, can be used to record the surfaces of test objects with a large number of measuring points. The object’s areas of interest must then be extracted from the acquired measurement data.