Vision-Based Measurement Method for Internal Riveting of the Joist

The requirement for industrial automation and quality control is the fundamental driver behind the visual examination of workpieces. Due to the vast number of workpieces, high level of complexity, and harsh production environment in traditional industrial production, manual inspection's efficiency and accuracy struggle to satisfy production standards. Automated vision inspection technology can partially address these issues. Through digital image processing and other technical methods, vision inspection technology can quickly and accurately identify and analyze features like workpiece size and shape, realizing automated quality control and production process control. In addition, workpiece vision inspection technology has grown in significance with the rise of Industry 4.0 and smart production. This paper's focus is the joist's rivet head, which must be measured for its diameter, height, circularity, and edge distance to assess whether the rivet is successfully assembled by confirming that the rivet head's dimensions are accurate. This paper suggests using Euclidean clustering to segment the point cloud surface and converting the extracted surface into an image for global threshold segmentation to extract edge information in an effort to overcome the challenges posed by the reflection and shadows on metal workpiece surfaces. This research suggests that the holes created by the blockage of light source irradiation on joist edges be filled by filling the pixel values. In this study, the riveted head's diameter is accurate to within 0.1 mm, its height is accurate to within 0.02 mm, and the edge distance is accurate to within 0.2 mm.