Indexed by:
Abstract:
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. © Springer Nature Switzerland AG 2018.
Keyword:
Reprint Author's Address:
Email:
Source :
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ISSN: 0302-9743
Year: 2018
Publish Date: 2018
Volume: 11164 LNCS
Page: 711-720
Language: English
0 . 4 0 2
JCR@2005
JCR Journal Grade:2
Cited Count:
WoS CC Cited Count: 4
SCOPUS Cited Count: 5
ESI Highly Cited Papers on the List: 0 Unfold All
WanFang Cited Count:
Chinese Cited Count:
30 Days PV: 21
Affiliated Colleges: