GPU assisted towards real-time reconstruction for dual-camera compressive hyperspectral imaging

Shipeng Zhang, Lizhi Wang, Ying Fu, Hua Huang*

*此作品的通讯作者

科研成果: 书/报告/会议事项章节会议稿件同行评审

5 引用 (Scopus)

摘要

The dual-camera compressive hyperspectral imager (DCCHI) can capture 3D hyperspectral image (HSI) with a single snapshot. However, due to the high computation complexity of reconstruction methods, DCCHI cannot apply to the time-crucial applications. In this paper, we propose a GPU assisted towards real-time reconstruction framework for DCCHI. First, leveraging the fast convergence rate of the alternative direction multiplier method, we propose a reformative reconstruction algorithm which can achieve a fast convergence rate. Then, using the interpolation results of a low resolution reconstructed HSI as the warm start, we propose a fast reconstruction strategy to further reduce the computation burden. Last, a GPU parallel implementation is presented to achieve nearly real-time reconstruction. Evaluation experiments indicate our framework can obtain a significant promotion in reconstruction efficiency with a slight accuracy loss.

源语言英语
主期刊名Advances in Multimedia Information Processing – PCM 2018 - 19th Pacific-Rim Conference on Multimedia, 2018, Proceedings
编辑Chong-Wah Ngo, Richang Hong, Meng Wang, Wen-Huang Cheng, Toshihiko Yamasaki
出版商Springer Verlag
711-720
页数10
ISBN(印刷版)9783030007751
DOI
出版状态已出版 - 2018
活动19th Pacific-Rim Conference on Multimedia, PCM 2018 - Hefei, 中国
期限: 21 9月 201822 9月 2018

出版系列

姓名Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
11164 LNCS
ISSN(印刷版)0302-9743
ISSN(电子版)1611-3349

会议

会议19th Pacific-Rim Conference on Multimedia, PCM 2018
国家/地区中国
Hefei
时期21/09/1822/09/18

指纹

探究 'GPU assisted towards real-time reconstruction for dual-camera compressive hyperspectral imaging' 的科研主题。它们共同构成独一无二的指纹。

引用此