arXiv subjects

Bardia Hejazi

Publications and source records attributed to Bardia Hejazi.

7 recordsLinked to original sources

arXiv2025

Deep Learning based 3D Volume Correlation for Additive Manufacturing Using High-Resolution Industrial X-ray Computed Tomography

Quality control in additive manufacturing (AM) is vital for industrial applications in areas such as the automotive, medical and aerospace sectors. Geometric inaccuracies caused by shrinkage and deformations can compromise the life and performance of additively manufactured components. Such deviations can be quantified using Digital Volume Correlation (DVC), which compares the computer-aided design (CAD) model with the X-ray Computed Tomography (XCT) geometry of the components produced. However, accurate registration between the two modalities is challenging due to the absence of a ground truth or reference deformation field. In addition, the extremely large data size of high-resolution XCT volumes makes computation difficult. In this work, we present a deep learning-based approach for estimating voxel-wise deformations between CAD and XCT volumes. Our method uses a dynamic patch-based processing strategy to handle high-resolution volumes. In addition to the Dice Score, we introduce a Binary Difference Map (BDM) that quantifies voxel-wise mismatches between binarized CAD and XCT volumes to evaluate the accuracy of the registration. Our approach shows a 9.2\% improvement in the Dice Score and a 9.9\% improvement in the voxel match rate compared to classic DVC methods, while reducing the interaction time from days to minutes. This work sets the foundation for deep learning-based DVC methods to generate compensation meshes that can then be used in closed-loop correlations during the AM production process. Such a system would be of great interest to industries since the manufacturing process will become more reliable and efficient, saving time and material.

cs.CV↗

arXiv2025

D-CNN and VQ-VAE Autoencoders for Compression and Denoising of Industrial X-ray Computed Tomography Images

The ever-growing volume of data in imaging sciences stemming from the advancements in imaging technologies, necessitates efficient and reliable storage solutions for such large datasets. This study investigates the compression of industrial X-ray computed tomography (XCT) data using deep learning autoencoders and examines how these compression algorithms affect the quality of the recovered data. Two network architectures with different compression rates were used, a deep convolution neural network (D-CNN) and a vector quantized variational autoencoder (VQ-VAE). The XCT data used was from a sandstone sample with a complex internal pore network. The quality of the decoded images obtained from the two different deep learning architectures with different compression rates were quantified and compared to the original input data. In addition, to improve image decoding quality metrics, we introduced a metric sensitive to edge preservation, which is crucial for three-dimensional data analysis. We showed that different architectures and compression rates are required depending on the specific characteristics needed to be preserved for later analysis. The findings presented here can aid scientists to determine the requirements and strategies for their data storage and analysis needs.

eess.IV↗

arXiv2022

Flight of a honeybee in turbulent wind

In windy conditions, the air is turbulent. The strong and intermittent velocity variations of turbulence are invisible to flying animals. Nevertheless, flying animals, not much larger than the smallest scales of turbulence, manage to maneuver these highly fluctuating conditions quite well. Here we quantify honeybee flight with time-resolved three-dimensional tracking in calm conditions and controlled turbulent winds. We find that honeybee mean speed and acceleration are only weakly correlated with the strength of turbulence. In flight, honeybees accelerate slowly and decelerate rapidly, i.e., they break suddenly during turns and then accelerate again. While this behavior is observed in both calm and turbulent conditions, it is increasingly dominant under turbulent conditions where short straight trajectories are broken by turns and increased maneuvering. This flight-crash behavior is reminiscent of turbulence itself. Our observations may help the development of flight strategies for miniature flying robotics under turbulent conditions.

physics.flu-dyn↗

arXiv2022

On the risk of infection by infectious aerosols in large indoor spaces

Airborne diseases can be transmitted by infectious aerosols in the near field, i.e., in close proximity, or in the far field, i.e., by infectious aerosols that are well mixed within the indoor air. Is it possible to say which mode of disease transmission is predominant in large indoor spaces? We addressed this question by measuring the transport of aerosols equivalent to the size of human respiratory particles in two large hardware stores (V>10000 m$^3$). We found that aerosol concentrations in both stores decreased rapidly and almost independently of aerosol size, despite the different ventilation systems. A persistent and directional airflow on the order of a few cm/s was observed in both stores. Consequently, aerosol dynamics in such open settings can be expected to be dominated by turbulent dispersion and sweeping, and the accumulation of infectious aerosols in the indoor air is unlikely to contribute significantly to the risk of infection as long as the occupancy of the store is not too high. Under these conditions, well-fitting face masks are an excellent means of preventing disease transmission by human aerosols.

q-bio.PE↗

arXiv2021

Face-masks save us from SARS-CoV-2 transmission

We present results on the infection risk from SARS-CoV-2 under different scenarios based on measured particle size-dependent mask penetration, measured total inward leakage, measured human aerosol emission for sizes from 10nm to 1mm, and re-hydration on inhalation. Well-mixed room models significantly underestimate the risk of infection for short and direct exposure. To this end, we estimate the upper bound for infection risk with the susceptible in the infectious exhalation cloud or wearing masks by having the masked susceptible inhale the entire exhalation of a masked infectious. Social distances without a mask, even at 3m between speaking individuals results in an upper bound of 90\% for risk of infection after a few minutes. If both wear a surgical mask, the risk of infection for the person speaking remains below 26\% even after 60 minutes. When both the infectious and susceptible wear a well-fitting FFP2 mask, the upper bound for risk is reduced by a factor of 60 compared to surgical masks. In both cases, face leakage is very important. For FFP2 masks, leakage is low in the nasal region and directed upward, which can be further reduced significantly by applying double-sided medical tape there. Considering that the calculated upper bound greatly overestimates the risk of infection, and the fact that with a poorly worn mask even the upper bound we calculated is very low, we conclude that wearing a mask, even with some leakage, provides excellent third party and self-protection.

physics.med-ph↗

arXiv2019

Using deformable particles for single particle measurements of velocity gradient tensors

We measure the deformation of particles made of several slender arms in a two-dimensional (2D) linear shear and a three-dimensional (3D) turbulent flow. We show how these measurements of arm deformations along with the rotation rate of the particle allow us to extract the velocity gradient tensor of the flow. The particles used in the experiments have three symmetric arms in a plane (triads) and are fabricated using 3D printing of a flexible polymeric material. Deformation measurements of a particle free to rotate about a fixed axis in a 2D simple shear flow are used to validate our model relating particle deformations to the fluid strain. We then examine deformable particles in a 3D turbulent flow created by a jet array in a vertical water tunnel. Particle orientations and deformations are measured with high precision using four high speed cameras and have an uncertainty on the order of $10^{-4}$ radians. Measured deformations in 3D turbulence are small and only slightly larger than our orientation measurement uncertainty. Simulation results for triads in turbulence show deformations similar to the experimental observations. Deformable particles offer a promising method for measuring the full local velocity gradient tensor from measurements of a single particle where traditionally a high concentration of tracer particles would be required.

cond-mat.soft↗

arXiv2017

Emergent scar lines in chaotic advection of passive directors

We examine the spatial field of orientations of slender fibers that are advected by a two-dimensional fluid flow. The orientation field of these passive directors are important in a wide range of industrial and geophysical flows. We introduce emergent scar lines as the dominant coherent structures in the orientation field of passive directors in chaotic flows. Previous work has identified the existence of scar lines where the orientation rotates by π over short distances, but the lines that were identified disappeared as time progressed. As a result, earlier work focused on topological singularities in the orientation field which we find to play a negligible role at long times. We use the standard map as a simple time-periodic two-dimensional (2D) flow that produces Lagrangian chaos. This class of flows produce persistent patterns in passive scalar advection, and we find that a different kind of persistent pattern develops in the passive director orientation field. We identify the mechanism by which emergent scar lines grow to dominate these patterns at long times in complex flows. Emergent scar lines form where the recent stretching of the fluid element is perpendicular to earlier stretching. Thus these scar lines can be labeled by their age, defined as the time since their stretching reached a maximum.

physics.flu-dyn↗