Segmentation and mixed reality representation of cavities in die cast aluminum using 3D x-ray CT

This study proposes a method for intuitively visualizing the three-dimensional structure of internal defects in aluminum die-cast parts imaged using industrial 3D x-ray CT in a Mixed Reality (MR) environment. Conventional CT visualization primarily employs cross-sectional images in 2D or surface/volume rendering in 3D for visualization. However, these methods have limitations in understanding the spatial structure of complex defect shapes. In this study, we developed an MR system that enables dynamic manipulation and visualization in response to the user's viewpoint and gestures, utilizing a spatial reality display and motion capture. 3D x-ray CT data is surface rendered using two thresholds (one for outer contours and one for casting cavities) and superimposed on Unity in a semi-transparent manner. Additionally, 2D cross-sectional images are generated in real-time from DICOM-format volume data and superimposed on the 3D model according to the user-specified cross-sectional position, enabling the integrated visualization of three-dimensional structures and local cross-sectional images. This enables the intuitive understanding of the penetration structure and spatial continuity of casting cavities, which were previously difficult to grasp using conventional methods. This method represents a novel spatial visualization approach for non-destructive testing (NDT) and holds significant potential for application across various industrial, medical, and security fields.