Show simple item record

dc.contributor.authorLi, Yuqi
dc.contributor.authorQi, Miao
dc.contributor.authorGulve, Rahul
dc.contributor.authorWei, Mian
dc.contributor.authorGenov, Roman
dc.contributor.authorKutulakos, Kiriakos N.
dc.contributor.authorHeidrich, Wolfgang
dc.date.accessioned2020-06-24T08:44:26Z
dc.date.available2020-06-24T08:44:26Z
dc.date.issued2020-06-02
dc.identifier.citationLi, Y., Qi, M., Gulve, R., Wei, M., Genov, R., Kutulakos, K. N., & Heidrich, W. (2020). End-to-End Video Compressive Sensing Using Anderson-Accelerated Unrolled Networks. 2020 IEEE International Conference on Computational Photography (ICCP). doi:10.1109/iccp48838.2020.9105237
dc.identifier.isbn978-1-7281-5231-8
dc.identifier.issn2164-9774
dc.identifier.doi10.1109/ICCP48838.2020.9105237
dc.identifier.urihttp://hdl.handle.net/10754/663826
dc.description.abstractCompressive imaging systems with spatial-temporal encoding can be used to capture and reconstruct fast-moving objects. The imaging quality highly depends on the choice of encoding masks and reconstruction methods. In this paper, we present a new network architecture to jointly design the encoding masks and the reconstruction method for compressive high-frame-rate imaging. Unlike previous works, the proposed method takes full advantage of denoising prior to provide a promising frame reconstruction. The network is also flexible enough to optimize full-resolution masks and efficient at reconstructing frames. To this end, we develop a new dense network architecture that embeds Anderson acceleration, known from numerical optimization, directly into the neural network architecture. Our experiments show the optimized masks and the dense accelerated network respectively achieve 1.5 dB and 1 dB improvements in PSNR without adding training parameters. The proposed method outperforms other state-of-the-art methods both in simulations and on real hardware. In addition, we set up a coded two-bucket camera for compressive high-frame-rate imaging, which is robust to imaging noise and provides promising results when recovering nearly 1,000 frames per second.
dc.publisherIEEE
dc.relation.urlhttps://ieeexplore.ieee.org/document/9105237/
dc.relation.urlhttps://ieeexplore.ieee.org/document/9105237/
dc.relation.urlhttps://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9105237
dc.rightsArchived with thanks to IEEE
dc.subjecthigh-frame-rate imaging
dc.subjectdeep neural network
dc.subjectcomputational camera
dc.titleEnd-to-End Video Compressive Sensing Using Anderson-Accelerated Unrolled Networks
dc.typeConference Paper
dc.contributor.departmentComputational Imaging Group
dc.contributor.departmentComputer Science Program
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
dc.contributor.departmentElectrical Engineering
dc.contributor.departmentVisual Computing Center (VCC)
dc.conference.date24-26 April 2020
dc.conference.name2020 IEEE International Conference on Computational Photography (ICCP)
dc.conference.locationSaint Louis, MO, USA
dc.eprint.versionPost-print
dc.contributor.institutionUniversity of Toronto,Canada
kaust.personLi, Yuqi
kaust.personQi, Miao
kaust.personHeidrich, Wolfgang
refterms.dateFOA2020-06-29T05:27:53Z
dc.date.published-online2020-06-02
dc.date.published-print2020-04


Files in this item

Thumbnail
Name:
main.pdf
Size:
2.917Mb
Format:
PDF
Description:
Accepted manuscript

This item appears in the following Collection(s)

Show simple item record