Abstract: This paper presents recent developments in statistical image processing to demonstrate feasibility of layer-by-layer nondestructive inspection for selective laser melting (SLM) of metallic parts. A matrix of 10x10mm Stainless 316 squares were deposited with the first row consisting of just the sintered surface, Row 2 a single layer, Row 3 with 2 layers and so on. These layers were scanned to emulate the layer-by-layer collection of top surface geometry using a laser sensor. The resultant data was utilized to generate ISO 25718 roughness parameters and subsequently to perform track identification. This calculation consisted of two salient steps: (1) Computing the gradients of surface roughness, which vanish at the peaks and pits and (2) Extracting the scanning direction by taking Hough Transform of the geometry. This algorithm was repeated for all layers and the alignment of the roughness with the rotating scan was observed to be persistent for all layers.
Authors: Subhadeep Chakraborty, Anil B. Chaudhary, and Sudarsanam S. Babu
Keywords: SLM, NDE, Hough Transform, Track Identification