{"id":{"repo_id":"gmu","oai_identifier":"oai:MARS:1920/13693"},"canonical_url":"https://search.dev.ndltd.org/etd/gmu/oai:MARS:1920/13693","repository":{"repo_id":"gmu","name":"George Mason University","base_url":"https://mars.gmu.edu/server/oai/request"},"display":{"title":"Neuromorphic Imaging: Modeling, Bilevel Optimization, and Generalized Nash Equilibrium","abstract":"This work delves into the innovative optimization algorithms and applications of event-based cameras in image segmentation, motion estimation, and image deblurring. Event-based cameras revolutionize visual information capture by employing a continuous stream of asynchronous events, departing from the traditional frame-based acquisition approach. Inspired by biological sensors, event-based cameras record data based on asynchronous changes in light intensity at each pixel. These cameras possess a temporal resolution of microseconds. The remarkable capability of neuromorphic cameras to capture fast-moving objects without motion blur provides a distinct advantage over traditional cameras that record synchronous average light intensity over an exposure time. Moreover, due to their asynchronous event-based nature, neuromorphic cameras are not susceptible to exposure-related issues commonly encountered in traditional cameras, further enhancing their imaging performance. We introduce a bilevel optimization framework specifically designed for neuromorphic imaging, which addresses the challenges associated with reconstructing high-quality images from event-based camera data. Under certain assumptions, we establish existence of a solution to the optimization problem and derive second-order sufficient conditions for this nonconvex problem. Additionally, we develop a second-order Newton-type solver to effectively tackle this complex problem. To further enhance the performance of our framework, we propose a learning inspired algorithm that computes optimal regularization parameters for each pixel individually. The effectiveness and versatility of our approach is demonstrated through a series of examples. To address the challenges in image segmentation and motion estimation, we introduce a Generalized Nash Equilibrium-based framework. This framework capitalizes on the temporal and spatial information derived from the event stream, enabling accurate segmentation and motion estimation. Theoretical foundations are established through the derivation of existence criteria and the proposal of a multi-level optimization method to calculate equilibrium. We demonstrate the effectiveness of our framework through a series of examples. By combining these research directions, this work offers a comprehensive exploration of event-based cameras' applications in image segmentation, motion estimation, and image deblurring within the realm of rigorous mathematics. The findings contribute to the field by introducing novel models, algorithms, analysis, and showcasing their effectiveness through several examples.","abstract_html":"This work delves into the innovative optimization algorithms and applications of event-based cameras in image segmentation, motion estimation, and image deblurring. Event-based cameras revolutionize visual information capture by employing a continuous stream of asynchronous events, departing from the traditional frame-based acquisition approach. Inspired by biological sensors, event-based cameras record data based on asynchronous changes in light intensity at each pixel. These cameras possess a temporal resolution of microseconds. The remarkable capability of neuromorphic cameras to capture fast-moving objects without motion blur provides a distinct advantage over traditional cameras that record synchronous average light intensity over an exposure time. Moreover, due to their asynchronous event-based nature, neuromorphic cameras are not susceptible to exposure-related issues commonly encountered in traditional cameras, further enhancing their imaging performance. We introduce a bilevel optimization framework specifically designed for neuromorphic imaging, which addresses the challenges associated with reconstructing high-quality images from event-based camera data. Under certain assumptions, we establish existence of a solution to the optimization problem and derive second-order sufficient conditions for this nonconvex problem. Additionally, we develop a second-order Newton-type solver to effectively tackle this complex problem. To further enhance the performance of our framework, we propose a learning inspired algorithm that computes optimal regularization parameters for each pixel individually. The effectiveness and versatility of our approach is demonstrated through a series of examples. To address the challenges in image segmentation and motion estimation, we introduce a Generalized Nash Equilibrium-based framework. This framework capitalizes on the temporal and spatial information derived from the event stream, enabling accurate segmentation and motion estimation. Theoretical foundations are established through the derivation of existence criteria and the proposal of a multi-level optimization method to calculate equilibrium. We demonstrate the effectiveness of our framework through a series of examples. By combining these research directions, this work offers a comprehensive exploration of event-based cameras' applications in image segmentation, motion estimation, and image deblurring within the realm of rigorous mathematics. The findings contribute to the field by introducing novel models, algorithms, analysis, and showcasing their effectiveness through several examples.","abstract_has_math":false,"creators":["Sayre, David Lee"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2023,"date_issued":"2023","date_published":"2023","updated_at":"2026-07-27T19:51:50Z","subjects":[],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["hdl:1920/13693"],"render_values":[{"text":"hdl:1920/13693","href":null,"code":true}]}]},"links":{"outbound_url":null,"outbound_label":null,"outbound_source":null},"metadata_groups":[{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2023"]},{"key":"dc:type","label":"Dc Type","values":["Dissertation"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["hdl:1920/13693"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.other","label":"Dc Description Other","values":["This work delves into the innovative optimization algorithms and applications of event-based cameras in image segmentation, motion estimation, and image deblurring. Event-based cameras revolutionize visual information capture by employing a continuous stream of asynchronous events, departing from the traditional frame-based acquisition approach. Inspired by biological sensors, event-based cameras record data based on asynchronous changes in light intensity at each pixel. These cameras possess a temporal resolution of microseconds. The remarkable capability of neuromorphic cameras to capture fast-moving objects without motion blur provides a distinct advantage over traditional cameras that record synchronous average light intensity over an exposure time. Moreover, due to their asynchronous event-based nature, neuromorphic cameras are not susceptible to exposure-related issues commonly encountered in traditional cameras, further enhancing their imaging performance. We introduce a bilevel optimization framework specifically designed for neuromorphic imaging, which addresses the challenges associated with reconstructing high-quality images from event-based camera data. Under certain assumptions, we establish existence of a solution to the optimization problem and derive second-order sufficient conditions for this nonconvex problem. Additionally, we develop a second-order Newton-type solver to effectively tackle this complex problem. To further enhance the performance of our framework, we propose a learning inspired algorithm that computes optimal regularization parameters for each pixel individually. The effectiveness and versatility of our approach is demonstrated through a series of examples. To address the challenges in image segmentation and motion estimation, we introduce a Generalized Nash Equilibrium-based framework. This framework capitalizes on the temporal and spatial information derived from the event stream, enabling accurate segmentation and motion estimation. Theoretical foundations are established through the derivation of existence criteria and the proposal of a multi-level optimization method to calculate equilibrium. We demonstrate the effectiveness of our framework through a series of examples. By combining these research directions, this work offers a comprehensive exploration of event-based cameras' applications in image segmentation, motion estimation, and image deblurring within the realm of rigorous mathematics. The findings contribute to the field by introducing novel models, algorithms, analysis, and showcasing their effectiveness through several examples."]},{"key":"dc:title","label":"Title","values":["Neuromorphic Imaging: Modeling, Bilevel Optimization, and Generalized Nash Equilibrium"]}]}],"canonical_facts":{"dc:date.issued":["2023"],"dc:description.other":["This work delves into the innovative optimization algorithms and applications of event-based cameras in image segmentation, motion estimation, and image deblurring. Event-based cameras revolutionize visual information capture by employing a continuous stream of asynchronous events, departing from the traditional frame-based acquisition approach. Inspired by biological sensors, event-based cameras record data based on asynchronous changes in light intensity at each pixel. These cameras possess a temporal resolution of microseconds. The remarkable capability of neuromorphic cameras to capture fast-moving objects without motion blur provides a distinct advantage over traditional cameras that record synchronous average light intensity over an exposure time. Moreover, due to their asynchronous event-based nature, neuromorphic cameras are not susceptible to exposure-related issues commonly encountered in traditional cameras, further enhancing their imaging performance. We introduce a bilevel optimization framework specifically designed for neuromorphic imaging, which addresses the challenges associated with reconstructing high-quality images from event-based camera data. Under certain assumptions, we establish existence of a solution to the optimization problem and derive second-order sufficient conditions for this nonconvex problem. Additionally, we develop a second-order Newton-type solver to effectively tackle this complex problem. To further enhance the performance of our framework, we propose a learning inspired algorithm that computes optimal regularization parameters for each pixel individually. The effectiveness and versatility of our approach is demonstrated through a series of examples. To address the challenges in image segmentation and motion estimation, we introduce a Generalized Nash Equilibrium-based framework. This framework capitalizes on the temporal and spatial information derived from the event stream, enabling accurate segmentation and motion estimation. Theoretical foundations are established through the derivation of existence criteria and the proposal of a multi-level optimization method to calculate equilibrium. We demonstrate the effectiveness of our framework through a series of examples. By combining these research directions, this work offers a comprehensive exploration of event-based cameras' applications in image segmentation, motion estimation, and image deblurring within the realm of rigorous mathematics. The findings contribute to the field by introducing novel models, algorithms, analysis, and showcasing their effectiveness through several examples."],"dc:identifier":["hdl:1920/13693"],"dc:title":["Neuromorphic Imaging: Modeling, Bilevel Optimization, and Generalized Nash Equilibrium"],"dc:type":["Dissertation"]},"updated_at":"2026-07-27T19:51:50Z"}