TY - GEN
T1 - GPU assisted towards real-time reconstruction for dual-camera compressive hyperspectral imaging
AU - Zhang, Shipeng
AU - Wang, Lizhi
AU - Fu, Ying
AU - Huang, Hua
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2018.
PY - 2018
Y1 - 2018
N2 - 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.
AB - 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.
KW - Alternative direction multiplier method
KW - Compressive hyperspectral imaging
KW - Parallel optimization
KW - Real-time reconstruction
UR - https://www.scopus.com/pages/publications/85057213862
U2 - 10.1007/978-3-030-00776-8_65
DO - 10.1007/978-3-030-00776-8_65
M3 - Conference contribution
AN - SCOPUS:85057213862
SN - 9783030007751
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 711
EP - 720
BT - Advances in Multimedia Information Processing – PCM 2018 - 19th Pacific-Rim Conference on Multimedia, 2018, Proceedings
A2 - Ngo, Chong-Wah
A2 - Hong, Richang
A2 - Wang, Meng
A2 - Cheng, Wen-Huang
A2 - Yamasaki, Toshihiko
PB - Springer Verlag
T2 - 19th Pacific-Rim Conference on Multimedia, PCM 2018
Y2 - 21 September 2018 through 22 September 2018
ER -