The widespread use of smartphones and camera-coupled Internet of Thing (IoT) devices triggers an explosive growth of imagery data. To extract and process the rich contents contained in imagery data, various image analysis techniques have been investigated and applied to a spectrum of application scenarios. In recent years, breakthroughs in deep learning have powered a new revolution for image analysis in terms of effectiveness with high resource consumption. Given the fact that most smartphones and IoT devices have limited computational capability and battery life, they are not ready for the processing of computational intensive analytics over imagery data collected by them, especially when deep learning is involved. To resolve the bottleneck of computation, storage, and energy for these resource constrained devices, offloading complex image analysis to public cloud computing platforms has become a promising trend in both academia and industry. However, an outstanding challenge with public cloud is on the protection of sensitive information contained in many imagery data, such as personal identities and financial data. Directly sending imagery data to the public cloud can cause serious privacy concerns and even legal issues.
In this dissertation, I propose a comprehensive privacy-preserving imagery data analysis framework which can be integrated in different application scenarios to assist image analysis for resource-constrained devices with efficiency, accuracy, and privacy protection. I first identify security challenges in the utilization of public cloud for image analysis. Then, I design and develop a set of novel solutions to address these challenges. These solutions will be featured by strong privacy guarantee, lightweight computation, low accuracy loss compared with image analysis without privacy protection. To optimize the communication overhead and resource utilization of using cloud computing, I investigate edge computing, which is a promising technique to ameliorate the high communication overhead in cloud-assisted architectures. Furthermore, to boost the performance of my solutions under both cloud and edge deployment, I also provide a set of pluggable enhancement modules to be applied to meet different requirements for various tasks. By exploring the features of edge computing and cloud computing, I flexibly incorporate them as a comprehensive framework to provide privacy-preserving image analysis services.
","abstract_html":"<p>The widespread use of smartphones and camera-coupled Internet of Thing (IoT) devices triggers an explosive growth of imagery data. To extract and process the rich contents contained in imagery data, various image analysis techniques have been investigated and applied to a spectrum of application scenarios. In recent years, breakthroughs in deep learning have powered a new revolution for image analysis in terms of effectiveness with high resource consumption. Given the fact that most smartphones and IoT devices have limited computational capability and battery life, they are not ready for the processing of computational intensive analytics over imagery data collected by them, especially when deep learning is involved. To resolve the bottleneck of computation, storage, and energy for these resource constrained devices, offloading complex image analysis to public cloud computing platforms has become a promising trend in both academia and industry. However, an outstanding challenge with public cloud is on the protection of sensitive information contained in many imagery data, such as personal identities and financial data. Directly sending imagery data to the public cloud can cause serious privacy concerns and even legal issues.</p> <p>In this dissertation, I propose a comprehensive privacy-preserving imagery data analysis framework which can be integrated in different application scenarios to assist image analysis for resource-constrained devices with efficiency, accuracy, and privacy protection. I first identify security challenges in the utilization of public cloud for image analysis. Then, I design and develop a set of novel solutions to address these challenges. These solutions will be featured by strong privacy guarantee, lightweight computation, low accuracy loss compared with image analysis without privacy protection. To optimize the communication overhead and resource utilization of using cloud computing, I investigate edge computing, which is a promising technique to ameliorate the high communication overhead in cloud-assisted architectures. Furthermore, to boost the performance of my solutions under both cloud and edge deployment, I also provide a set of pluggable enhancement modules to be applied to meet different requirements for various tasks. By exploring the features of edge computing and cloud computing, I flexibly incorporate them as a comprehensive framework to provide privacy-preserving image analysis services.</p>","abstract_has_math":false,"creators":["Tian, Yifan"],"institution":null,"degree_name":"Doctor of Philosophy in Electrical Engineering & Computer Science","degree_level":"Dissertation - Open Access","degree_discipline":"Electrical, Computer, Software, and Systems Engineering","degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2019,"date_issued":"2019-11-01T07:00:00Z","date_published":"2019-11-01T07:00:00Z","updated_at":"2026-07-27T19:25:37Z","subjects":["data analysis","privacy","imagery","Electrical and Computer Engineering","Privacy Law"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://commons.erau.edu/edt/480","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Tian, Yifan"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"thesis:degree_discipline","label":"Discipline","values":["Electrical, Computer, Software, and Systems Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Dissertation - Open Access"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy in Electrical Engineering & Computer Science"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["data analysis","privacy","imagery","Electrical and Computer Engineering","Privacy Law"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://commons.erau.edu/edt/480"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The widespread use of smartphones and camera-coupled Internet of Thing (IoT) devices triggers an explosive growth of imagery data. To extract and process the rich contents contained in imagery data, various image analysis techniques have been investigated and applied to a spectrum of application scenarios. In recent years, breakthroughs in deep learning have powered a new revolution for image analysis in terms of effectiveness with high resource consumption. Given the fact that most smartphones and IoT devices have limited computational capability and battery life, they are not ready for the processing of computational intensive analytics over imagery data collected by them, especially when deep learning is involved. To resolve the bottleneck of computation, storage, and energy for these resource constrained devices, offloading complex image analysis to public cloud computing platforms has become a promising trend in both academia and industry. However, an outstanding challenge with public cloud is on the protection of sensitive information contained in many imagery data, such as personal identities and financial data. Directly sending imagery data to the public cloud can cause serious privacy concerns and even legal issues.
In this dissertation, I propose a comprehensive privacy-preserving imagery data analysis framework which can be integrated in different application scenarios to assist image analysis for resource-constrained devices with efficiency, accuracy, and privacy protection. I first identify security challenges in the utilization of public cloud for image analysis. Then, I design and develop a set of novel solutions to address these challenges. These solutions will be featured by strong privacy guarantee, lightweight computation, low accuracy loss compared with image analysis without privacy protection. To optimize the communication overhead and resource utilization of using cloud computing, I investigate edge computing, which is a promising technique to ameliorate the high communication overhead in cloud-assisted architectures. Furthermore, to boost the performance of my solutions under both cloud and edge deployment, I also provide a set of pluggable enhancement modules to be applied to meet different requirements for various tasks. By exploring the features of edge computing and cloud computing, I flexibly incorporate them as a comprehensive framework to provide privacy-preserving image analysis services.
"]},{"key":"dc:title","label":"Title","values":["Efficient Privacy-Aware Imagery Data Analysis"]}]}],"canonical_facts":{"dc:creator":["Tian, Yifan"],"dc:description.abstract":["The widespread use of smartphones and camera-coupled Internet of Thing (IoT) devices triggers an explosive growth of imagery data. To extract and process the rich contents contained in imagery data, various image analysis techniques have been investigated and applied to a spectrum of application scenarios. In recent years, breakthroughs in deep learning have powered a new revolution for image analysis in terms of effectiveness with high resource consumption. Given the fact that most smartphones and IoT devices have limited computational capability and battery life, they are not ready for the processing of computational intensive analytics over imagery data collected by them, especially when deep learning is involved. To resolve the bottleneck of computation, storage, and energy for these resource constrained devices, offloading complex image analysis to public cloud computing platforms has become a promising trend in both academia and industry. However, an outstanding challenge with public cloud is on the protection of sensitive information contained in many imagery data, such as personal identities and financial data. Directly sending imagery data to the public cloud can cause serious privacy concerns and even legal issues.
In this dissertation, I propose a comprehensive privacy-preserving imagery data analysis framework which can be integrated in different application scenarios to assist image analysis for resource-constrained devices with efficiency, accuracy, and privacy protection. I first identify security challenges in the utilization of public cloud for image analysis. Then, I design and develop a set of novel solutions to address these challenges. These solutions will be featured by strong privacy guarantee, lightweight computation, low accuracy loss compared with image analysis without privacy protection. To optimize the communication overhead and resource utilization of using cloud computing, I investigate edge computing, which is a promising technique to ameliorate the high communication overhead in cloud-assisted architectures. Furthermore, to boost the performance of my solutions under both cloud and edge deployment, I also provide a set of pluggable enhancement modules to be applied to meet different requirements for various tasks. By exploring the features of edge computing and cloud computing, I flexibly incorporate them as a comprehensive framework to provide privacy-preserving image analysis services.
"],"dc:identifier":["https://commons.erau.edu/edt/480"],"dc:subject":["data analysis","privacy","imagery","Electrical and Computer Engineering","Privacy Law"],"dc:title":["Efficient Privacy-Aware Imagery Data Analysis"],"thesis:degree_discipline":["Electrical, Computer, Software, and Systems Engineering"],"thesis:degree_level":["Dissertation - Open Access"],"thesis:degree_name":["Doctor of Philosophy in Electrical Engineering & Computer Science"]},"updated_at":"2026-07-27T19:25:37Z"}