Multiple-object segmentation by using a destructive 3D digitizer

Abstract

Abstract A destructive 3D digitizer is a practical tool in reverse engineering. It acquires images of the cross-sections of an object layer by layer. Then the desired portions of the images will be segmented to reconstruct a CAD model. While the threshold method was helpful to segment images containing a single object, it led to a substantial error for images containing multiple objects due to the reflection between objects. This study incorporated the least square method into image processing to correct reflection error for multiple-object reconstruction. As illustrated in the case study, we successfully removed the material reflection errors, which ranged from 4.5% to 19% of the total thickness of the reconstructed CAD model.