Layered Image Collection for Real-Time Defective Inspection in Additive Manufacturing

Abstract In Additive Manufacturing (AM), detecting cyber-attacks on infill structure is difficult because interior defects can occur without affecting the exterior. To detect the infill defectives quickly, layer-by-layer image inspection in real-time can be conducted. However, collecting the layered images from the top view in real-time is challenging because the 3D printer’s extruder interferes with objects from being perfectly scanned. Using a dummy model to move the extruder out of the object’s layer has been proposed. However, it is not practical because it creates printing delays and wasted printing materials. To enable infill layered image collection in real-time without delays and material waste, this research proposes a layered image collection method using an algorithm identifying a pseudo area in a layered image. The algorithm detects the pseudo area — the area covered by the extruder — using an image processing technique, such as an average pooling and max pooling. It accumulates the non-pseudo areas until a complete layered image is acquired. To validate and evaluate the proposed method, captured images were evaluated with various machine learning algorithms.