A 3D Machine Vision-Based Inspection System for Large-Scale Objects

The inspection of large-scale and complex objects is still a significant challenge in both academic and industrial fields. The main reason is that the large-scale and complex object, required to inspect from multiple views and positions, which makes it difficult to capture and analyze 3D data accurately. To solve this problems, this paper presents a novel inspection system for large-scale and complex objects based on 3D machine vision system, which includes: (1) a new algorithm for segmenting and recognizing 2D images and 3D point clouds; (2) a new inspection algorithm for large-scale and complex objects. Experimental results demonstrated that the proposed method significantly improves accuracy and computational efficiency for the inspection of large-scale and complex objects. In addition, this method has been successfully applied to an industrial inspection project in a factory of Volkswagen group.