Implementation of a combined multi-view laser line 3D reconstruction system

The Inline Computational Imaging (ICI) technique developed by the Competence Unit High-Performance Vision Systems (HVS) of the Austrian Institute of Technology (AIT) enables simultaneous 2D and 3D quality inspection of manufactured parts using a single image sensor technology and advanced reconstruction algorithms. The ICI pipeline reconstructs objects in 3D by leveraging multi-view stereo algorithms that identify common features in multiple images. However, this approach is limited to objects with surface texture, such as banknotes or steel billets. Objects with smooth surfaces, like keys, pose a challenge due to the lack of inherent texture at the resolutions used for analysis. To address this limitation, this master’s thesis project aims to enhance the ICI setup by incorporating an additional laser line illumination, creating a "Laser-augmented ICI" system. When the laser line interacts with the object’s surface, it bends based on the local surface curvature. By capturing the distortion of the laser line using the camera’s sensor, spatially resolved depth information can be inferred, even for objects without inherent surface texture. Integrating this depth cue into the existing ICI reconstruction algorithms would enable more reliable inspection of objects lacking significant surface texture relevant to the inspection task.

The main goal of this master’s thesis is to investigate and evaluate the potential improvement in 3D reconstruction quality achievable with the "Laser-augmented ICI" compared to standard ICI. The research hypothesis to be explored is whether the additional laser line illumination and the derived depth cues can enhance the accuracy of 3D reconstruction obtained from the acquired data.