About this Abstract |
Meeting |
2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024)
|
Symposium
|
2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024)
|
Presentation Title |
Gradient Segmentation of In-Situ Infrared Images for Porosity Detection in Electron Beam Powder Bed Fusion |
Author(s) |
Brian Johnstone, Christopher Saldana |
On-Site Speaker (Planned) |
Brian Johnstone |
Abstract Scope |
For metal additive manufacturing to be more effectively and widely used, greater process control is needed. One way to achieve this is through in-situ process monitoring, such as using layer-wise infrared imaging to detect porosity in electron beam powder bed fusion. Due to the pores having more emissivity than the solid part, they appear brighter in infrared images and can therefore be detected via image processing techniques. This work compares how applying different image filtering and gradient types can detect these brighter spots correlating to developing pores. Results were assessed both qualitatively via image appearance and histogram distributions and quantitatively via X-ray computed tomography scans. Gradients that use larger kernel sizes (specifically three-by-three) were more accurate in detecting porosity, and this was further aided by an anisotropic diffusion filter. This work provides objective insight into using gradient-based segmentation for academic and industry defect detection for greater process control. |
Proceedings Inclusion? |
Definite: Post-meeting proceedings |