{"id":{"repo_id":"csusb","oai_identifier":"oai:scholarworks.lib.csusb.edu:etd-3340"},"canonical_url":"https://search.ndltd.org/etd/csusb/oai:scholarworks.lib.csusb.edu:etd-3340","repository":{"repo_id":"csusb","name":"CSUniversity San Bernardino","base_url":"https://scholarworks.lib.csusb.edu/do/oai/"},"display":{"title":"APPLICATIONS OF THE FREQUENCY DOMAIN IN AUTONOMOUS DRIVING","abstract":"This thesis explores applications of the frequency domain (also known as spectral domain) in autonomous driving. Through the application of the discrete Fourier transform to images and features used by autonomous driving systems, an invaluable alternative representation of this information is achieved. In this domain, any pointwise modification of data globally impacts the spatial domain, making global receptive fields easy to obtain. Furthermore, by considering the constituent frequencies of the data in this domain, it is possible to separate and focus on the low and high frequency components of the data, yielding new insights into its structure. In this work, the value of this approach will be demonstrated empirically. A novel data augmentation strategy exploiting frequency domain representations will be shown to significantly improve the accuracy of various state-of-the-art (SOTA) Bird's-Eye-View map segmentation networks. Applying the frequency domain directly within the network itself proves valuable as well, through the application of Fast Fourier Convolution to the BEV features. These improvements are found when applying Fast Fourier Convolution to both BEVDet based models and transformer based ones such as MapTR. These results show the underutilized potential of frequency domain representations to enhance autonomous driving perception systems. Due to the broad applicability of the Fourier transform to various data types, both discrete and continuous, there remains a great untapped potential to exploit the frequency domain in computer vision. It is hoped that this work will provide new insights to the research community and encourage further explorations of the frequency domain in autonomous driving.","abstract_html":"This thesis explores applications of the frequency domain (also known as spectral domain) in autonomous driving. Through the application of the discrete Fourier transform to images and features used by autonomous driving systems, an invaluable alternative representation of this information is achieved. In this domain, any pointwise modification of data globally impacts the spatial domain, making global receptive fields easy to obtain. Furthermore, by considering the constituent frequencies of the data in this domain, it is possible to separate and focus on the low and high frequency components of the data, yielding new insights into its structure. In this work, the value of this approach will be demonstrated empirically. A novel data augmentation strategy exploiting frequency domain representations will be shown to significantly improve the accuracy of various state-of-the-art (SOTA) Bird&#x27;s-Eye-View map segmentation networks. Applying the frequency domain directly within the network itself proves valuable as well, through the application of Fast Fourier Convolution to the BEV features. These improvements are found when applying Fast Fourier Convolution to both BEVDet based models and transformer based ones such as MapTR. These results show the underutilized potential of frequency domain representations to enhance autonomous driving perception systems. Due to the broad applicability of the Fourier transform to various data types, both discrete and continuous, there remains a great untapped potential to exploit the frequency domain in computer vision. It is hoped that this work will provide new insights to the research community and encourage further explorations of the frequency domain in autonomous driving.","abstract_has_math":false,"creators":["Glisson, Calvin A"],"institution":null,"degree_name":"Master of Science in Computer Science","degree_level":"Restricted Thesis: Campus only access","degree_discipline":"School of Computer Science and Engineering","degree_department":null,"school":null,"contributors":["Chen, Qiuxiao"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-11-19T08:09:58Z","subjects":["machine learning","artificial intelligence","neural networks","computer vision","AI","ML","Artificial Intelligence and Robotics","Data Science","Signal Processing"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":"https://scholarworks.lib.csusb.edu/etd/2147","outbound_label":"Repository record","outbound_source":"dc:identifier","record_url":"https://scholarworks.lib.csusb.edu/etd/2147","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://scholarworks.lib.csusb.edu/do/oai/?verb=GetRecord&metadataPrefix=oai_etdms&identifier=oai%3Ascholarworks.lib.csusb.edu%3Aetd-3340","prefix":"oai_etdms"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Chen, Qiuxiao"]},{"key":"dc:creator","label":"Author","values":["Glisson, Calvin A"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2025-04-03T07:00:00Z"]},{"key":"dc:date.created","label":"Dc Date Created","values":["2025-05-01T07:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["School of Computer Science and Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Restricted Thesis: Campus only access"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science in Computer Science"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["machine learning","artificial intelligence","neural networks","computer vision","AI","ML","Artificial Intelligence and Robotics","Data Science","Signal Processing"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://scholarworks.lib.csusb.edu/etd/2147"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["This thesis explores applications of the frequency domain (also known as spectral domain) in autonomous driving. Through the application of the discrete Fourier transform to images and features used by autonomous driving systems, an invaluable alternative representation of this information is achieved. In this domain, any pointwise modification of data globally impacts the spatial domain, making global receptive fields easy to obtain. Furthermore, by considering the constituent frequencies of the data in this domain, it is possible to separate and focus on the low and high frequency components of the data, yielding new insights into its structure. In this work, the value of this approach will be demonstrated empirically. A novel data augmentation strategy exploiting frequency domain representations will be shown to significantly improve the accuracy of various state-of-the-art (SOTA) Bird's-Eye-View map segmentation networks. Applying the frequency domain directly within the network itself proves valuable as well, through the application of Fast Fourier Convolution to the BEV features. These improvements are found when applying Fast Fourier Convolution to both BEVDet based models and transformer based ones such as MapTR. These results show the underutilized potential of frequency domain representations to enhance autonomous driving perception systems. Due to the broad applicability of the Fourier transform to various data types, both discrete and continuous, there remains a great untapped potential to exploit the frequency domain in computer vision. It is hoped that this work will provide new insights to the research community and encourage further explorations of the frequency domain in autonomous driving."]},{"key":"dc:title","label":"Title","values":["APPLICATIONS OF THE FREQUENCY DOMAIN IN AUTONOMOUS DRIVING"]}]}],"canonical_facts":{"dc:contributor":["Chen, Qiuxiao"],"dc:creator":["Glisson, Calvin A"],"dc:date.available":["2025-04-03T07:00:00Z"],"dc:date.created":["2025-05-01T07:00:00Z"],"dc:description.abstract":["This thesis explores applications of the frequency domain (also known as spectral domain) in autonomous driving. Through the application of the discrete Fourier transform to images and features used by autonomous driving systems, an invaluable alternative representation of this information is achieved. In this domain, any pointwise modification of data globally impacts the spatial domain, making global receptive fields easy to obtain. Furthermore, by considering the constituent frequencies of the data in this domain, it is possible to separate and focus on the low and high frequency components of the data, yielding new insights into its structure. In this work, the value of this approach will be demonstrated empirically. A novel data augmentation strategy exploiting frequency domain representations will be shown to significantly improve the accuracy of various state-of-the-art (SOTA) Bird's-Eye-View map segmentation networks. Applying the frequency domain directly within the network itself proves valuable as well, through the application of Fast Fourier Convolution to the BEV features. These improvements are found when applying Fast Fourier Convolution to both BEVDet based models and transformer based ones such as MapTR. These results show the underutilized potential of frequency domain representations to enhance autonomous driving perception systems. Due to the broad applicability of the Fourier transform to various data types, both discrete and continuous, there remains a great untapped potential to exploit the frequency domain in computer vision. It is hoped that this work will provide new insights to the research community and encourage further explorations of the frequency domain in autonomous driving."],"dc:identifier":["https://scholarworks.lib.csusb.edu/etd/2147"],"dc:subject":["machine learning","artificial intelligence","neural networks","computer vision","AI","ML","Artificial Intelligence and Robotics","Data Science","Signal Processing"],"dc:title":["APPLICATIONS OF THE FREQUENCY DOMAIN IN AUTONOMOUS DRIVING"],"thesis:degree_discipline":["School of Computer Science and Engineering"],"thesis:degree_level":["Restricted Thesis: Campus only access"],"thesis:degree_name":["Master of Science in Computer Science"]},"updated_at":"2026-11-19T08:09:58Z"}