arXiv subjects

Kyle Williams

Publications and source records attributed to Kyle Williams.

4 recordsLinked to original sources

arXiv2026

Optimization of Precipitate Segmentation Through Linear Genetic Programming of Image Processing

Current analysis of additive manufactured niobium-based copper alloys relies on hand annotation due to varying contrast, noise, and image artifacts present in micrographs, slowing iteration speed in alloy development. We present a filtering and segmentation algorithm for detecting precipitates in FIB cross-section micrographs, optimized using linear genetic programming (LGP), which accounts for the various artifacts. To this end, the optimization environment uses a domain-specific language for image processing to iterate on solutions. Programs in this language are a list of image-filtering blocks with tunable parameters that sequentially process an input image, allowing for reliable generation and mutation by a genetic algorithm. Our environment produces optimized human-interpretable MATLAB code representing an image filtering pipeline. Under ideal conditions--a population size of 60 and a maximum program length of 5 blocks--our system was able to find a near-human accuracy solution with an average evaluation error of 1.8% when comparing segmentations pixel-by-pixel to a human baseline using an XOR error evaluation. Our automation work enabled faster iteration cycles and furthered exploration of the material composition and processing space: our optimized pipeline algorithm processes a 3.6 megapixel image in about 2 seconds on average. This ultimately enables convergence on strong, low-activation, precipitation hardened copper alloys for additive manufactured fusion reactor parts.

cs.CV↗

arXiv2024

Automated Registration of 3D Neurovascular Territory Atlas to 2D DSA for Targeted Quantitative Angiography Analysis

Subarachnoid hemorrhage (SAH), typically due to intracranial aneurysms, demands precise imaging for effective treatment. Digital Subtraction Angiography (DSA), despite being the gold standard, broadly visualizes cerebral blood flow, potentially masking key details in areas. This study introduces an approach integrating a 3D vascular atlas with 2D DSA images to allow targeted quantitative analysis in these crucial regions, thus enhancing diagnostic accuracy during interventions. Initially, DSA data was examined to ascertain the injection site. Following this, the appropriate viewing angle was determined to align accurately with the 3D vascular atlas. Utilizing this atlas, regions corresponding to the areas indicated as perfused were selected. Concurrently, a mask representing the perfused areas was created from the DSA sequence. This mask facilitated the initial coarse alignment of the projected 3D atlas to the DSA perfused territory deformable registration techniques, ensuring a precise overlay with the DSAs perfused territories. The performance of each overlay was measured using the Structural Similarity Index Measure (SSIM). The coregistration process revealed that deformable registrations was essential to achieve precise overlays of the 3D atlas projections with the 2D DSA perfused areas. This approach enabled the extraction of targeted quantitative angiography parameters, essential for detailed vascular assessment in subarachnoid hemorrhage cases. The integration of 3D atlas registration with 2D DSA projections facilitates a more precise and targeted diagnostic process for SAH during critical interventions. This image processing strategy enhances the visualization of affected arterial territories, potentially improving the accuracy of diagnostics and supporting better informed clinical decisions at the time of intervention

eess.IV↗

arXiv2021

Efficient Evader Detection in Mobile Sensor Networks

Suppose one wants to monitor a domain with sensors, each sensing a small ball-shaped region, but the domain is hazardous enough that one cannot control the placement of the sensors. A prohibitively large number of randomly placed sensors could be required to obtain static coverage. Instead, one can use fewer sensors by providing mobile coverage, a generalization of the static setup wherein every possible evader is detected by the moving sensors in a bounded amount of time. Here, we use topology in order to implement algorithms certifying mobile coverage that use only local data to solve the global problem. Our algorithms do not require knowledge of the sensors' locations, only their connectivity information. We experimentally study the statistics of mobile coverage in two dynamical scenarios. We allow the sensors to move independently (billiard dynamics and Brownian motion), or to locally coordinate their dynamics (collective animal motion models). Our detailed simulations show, for example, that collective motion can enhance performance: The expected time until the mobile sensor network achieves mobile coverage is lower for the D'Orsogna collective motion model than for the billiard motion model. Further, we show that even when the probability of static coverage is low, all possible evaders can nevertheless be detected relatively quickly by the mobile sensor network.

math.DS↗

arXiv2018

Real-Time Stochastic Predictive Control for Hybrid Vehicle Energy Management

This work presents three computational methods for real time energy management in a hybrid hydraulic vehicle (HHV) when driver behavior and vehicle route are not known in advance. These methods, implemented in a receding horizon control (aka model predictive control) framework, are rather general and can be applied to systems with nonlinear dynamics subject to a Markov disturbance. State and input constraints are considered in each method. A mechanism based on the steady state distribution of the underlying Markov chain is developed for planning beyond a finite horizon in the HHV energy management problem. Road elevation information is forecasted along the horizon and then merged with the statistical model of driver behavior to increase accuracy of the horizon optimization. The characteristics of each strategy are compared and the benefit of learning driver behavior is analyzed through simulation on three drive cycles, including one real world drive cycle. A simulation is designed to explicitly demonstrate the benefit of adapting the Markov chain to real time driver behavior. Experimental results demonstrate the real time potential of the primary algorithm when implemented on a processor with limited computational resources.

eess.SY↗