Bimodal Visualization of Industrial X-Ray and Neutron Computed Tomography Data

TL;DR

The paper addresses the challenge of visualizing industrial objects scanned with both X-ray and neutron CT by introducing a semiautomated bimodal visualization pipeline. It couples a Morse-complex segmentation of the joint bivariate histogram with an interactive histogram-to-color mapping and a real-time, co-registered bimodal renderer based on OSPRay. Key contributions include the relevance-based segmentation metric, an intuitive painting widget for refinement, and four case studies with expert feedback demonstrating efficient material identification and exploration. The approach yields a fast, understandable overview of multimaterial structures while preserving data fidelity, and is released as open-source to enable broader adoption and extension in nondestructive evaluation workflows.

Abstract

Advanced manufacturing creates increasingly complex objects with material compositions that are often difficult to characterize by a single modality. Our collaborating domain scientists are going beyond traditional methods by employing both X-ray and neutron computed tomography to obtain complementary representations expected to better resolve material boundaries. However, the use of two modalities creates its own challenges for visualization, requiring either complex adjustments of bimodal transfer functions or the need for multiple views. Together with experts in nondestructive evaluation, we designed a novel interactive bimodal visualization approach to create a combined view of the co-registered X-ray and neutron acquisitions of industrial objects. Using an automatic topological segmentation of the bivariate histogram of X-ray and neutron values as a starting point, the system provides a simple yet effective interface to easily create, explore, and adjust a bimodal visualization. We propose a widget with simple brushing interactions that enables the user to quickly correct the segmented histogram results. Our semiautomated system enables domain experts to intuitively explore large bimodal datasets without the need for either advanced segmentation algorithms or knowledge of visualization techniques. We demonstrate our approach using synthetic examp

Figures (15)

• Figure 1: The 1D histogram of the X-ray (blue) and neutron channel (green) of JH2B, which has five materials. The distinct peaks differ greatly, showing that one modality is inherently insufficient to capture all materials. In the X-ray, we can see three distinct peaks, whereas the neuron has three lower density peaks and a wider range of values that are spread out without any significant additional peak.
• Figure 2: Pipeline overview. We load the X-ray and neutron data at full resolution, generate the bivariate histogram, and compute the Morse-complex segmentation. Both the segmentation and the dataset itself are set as input for the renderer. During the ray casting process, when a voxel is queried it will be sampled for both modalities, resulting in a pair of sampled values. The 2D vector corresponds to the 2D position in the bivariate histogram. The RGB values are determined by the same coordinate lookup in the segmentation image, which is then modifiable through the application's built-in GUI.
• Figure 3: A Test example: The Simulated Synthetic Cylinder that contains four concentric cylinders of different combinations of values from the X-ray and the neutron channel. (a) Shows the ground truth, (b) and (c) show a slice of the two channels respectively, with X-ray missing the innermost hole and Neutron falsely mixing orange and green materials. (d) Shows that the materials and backgrounds are well separated into 4+2 clusters on the bivariate histogram on the left. The information is successfully captured in the Morse-complex segmentation on the right.
• Figure 4: Persistence vs. relevance simplification in a 1D histogram. (a) Original histogram segmented into modes indicated by color with the relevant persistence and relevance values of the local maxima indicated. (b) Hierarchical simplification by increasing persistence, which prioritizes high value peaks corresponding to common values. (c) Hierarchical simplification by decreasing relevance, which prioritizes distinct peaks when compared to their local neighborhood and preserves rare values.
• Figure 5: JH2B bivariate histogram (a) segmented with persistence (b) and relevance (c). With persistence, the purple color segment in (b) tries to further split the highest value peak into yellow and pink segments indicated by arrows, whereas the relevance imposes a better segment-material correlation by identifying a possible new region around the lower value area (brown, light blue, and yellow in (c)). (d)-(f) shows the ground truth illustration and the volume result of segmenting by persistence vs by relevance. The persistence scheme tends to segment out the bigger volume with smooth value variation, whereas relevance prioritized the smaller pieces with more distinct value peaks that are more likely corresponding to individual materials.