近年來，太赫茲輻射因其獨特的特性引起了各個研究領域的極大興趣。太赫茲輻射可以穿透非導電材料，並在照射不同物質時表現出獨特的反應。與X射線不同，太赫茲輻射由於光子能量低，對活體組織和DNA相對溫和。因此，太赫茲輻射被廣泛認為是電路缺陷檢測、材料辨識、醫學影像等應用的優秀工具和技術。
雖然太赫茲輻射具有巨大的發展潛力，但在實際應用上仍存在挑戰。例如，太赫茲輻射源產生器和接收器的成本高昂，使得使用像素陣列成像不可行。此外，依賴單一接收器的光柵掃描方法由於取樣時間較長，仍有改進的空間。因此，太赫茲成像的主流方法是單像素壓縮感測成像系統。利用訊號稀疏性，單像素成像系統可以以低於奈奎斯特取樣率的樣本計數運行，並使用壓縮感測理論重建訊號，從而顯著減少所需的取樣持續時間。
近年來，時間分辨太赫茲光譜成像已成為材料辨識和非侵入性成像的強大工具。然而，該技術普遍遇到了資料採集時間延長的挑戰。同時，由於資料量龐大，使用現有演算法重建完整的太赫茲脈衝資訊需要過長的持續時間。
本研究提出了一種用於太赫茲單像素成像系統的新型基於三維張量的壓縮感測模型。與僅使用峰值進行影像圖案分析的傳統系統不同，該系統可用於導出完整的太赫茲脈衝資訊以進行折射率分析。 由於此三維模型使用張量壓縮感測格式壓縮空間和時間域中的太赫茲響應，因此它大大縮短了訊號重建持續時間，同時加快了材料識別過程。 在演算法方面，本研究引入了張量廣義迭代軟閾值（T-GIST）演算法。與現有的張量壓縮感知恢復演算法相比，該演算法表現出優越的重建品質和速度性能。例如，與N-BOMP相比，T-GIST演算法的運行時間加快了6.34倍，重建品質提高了35％。對於GTCS-P演算法，所提出的T-GIST演算法將運行時間加快了67.8倍，並將重建品質提高了21%。
此外，本研究提出了T-GIST演算法的硬體架構，並透過在 Xilinx ZCU102 MPSoC 平台上的實作對其進行了驗證。實作結果表明，所提出的硬體架構只需0.013秒即可完成重建過程。

Terahertz radiation has attracted significant interest across various research domains in recent years due to its distinctive properties. Terahertz radiation can penetrate non-conductive materials and displays unique responses when irradiating different substances. Unlike X-rays, terahertz radiation is gentle on living tissues and DNA because of its low photon energy. Therefore, terahertz radiation is widely recognized as an excellent tool and technique for applications in circuit defect detection, material identification, medical imaging, etc.
While terahertz radiation holds substantial potential for development, there remain challenges in practical applications. For instance, the costliness of source generators and receivers for terahertz radiation renders imaging with pixel arrays unfeasible. Additionally, the method of raster scanning, reliant on a single receiver, still has room for improvement due to its extended sampling time. Consequently, the prevailing approach in terahertz imaging is the single-pixel compressive sensing imaging system. Leveraging signal sparsity, the single-pixel imaging system can operate with a sample count below the Nyquist sampling rate and reconstruct the signal using compressive sensing theory, significantly reducing the required sampling duration.
In recent years, time-resolved terahertz spectroscopic imaging has emerged as a powerful tool for material identification and non-invasive imaging. However, this technique generally encountered the challenge of the extended data acquisition time. Simultaneously, reconstructing complete terahertz pulse information using existing algorithms requires an excessive temporal duration due to the substantial volume of data.
This study proposes a novel 3D tensor-based compressive sensing model for the terahertz single-pixel imaging system. Unlike traditional systems that only use peak values for image pattern analysis, this proposed system can be used to derive the complete terahertz pulse information for refraction index analysis. As this 3-D model compresses the terahertz response in both spatial and time domains using the tensor compressive sensing format, it considerably diminished the signal reconstruction duration while expediting material identification processes. Concerning the algorithm, this study introduces the tensor generalized iterative soft thresholding (T-GIST) algorithm. Compared to existing tensor compressive sensing restoration algorithms, the proposed algorithm demonstrates superior reconstruction quality and speed performance. For instance, compared with the N-BOMP, the T-GIST algorithm accelerates the elapsed time by 6.34 times and improves the reconstruction quality by 35\%. For the GTCS-P algorithm, the proposed T-GIST algorithm speeds up 67.8 times the elapsed time and improves the reconstruction quality by 21\%.
Furthermore, this study proposes the hardware architecture for the T-GIST algorithm and validates it through implementation on the Xilinx ZCU102 MPSoC platform. The implementation results demonstrate the proposed hardware architecture accomplishes the reconstruction process in just 0.013 seconds.

[1] P.H. Siegel. Terahertz technology. IEEE Transactions on Microwave Theory and
Techniques, 50(3):910–928, 2002.
[2] Ashish Y. Pawar, Deepak D. Sonawane, Kiran B. Erande, and Deelip V. Derle.
Terahertz technology and its applications. Drug Invention Today, 5(2):157–163,
2013.
[3] Toshihiko Kiwa, Masayoshi Tonouchi, Masatsugu Yamashita, and Kodo Kawase.
Laser terahertz-emission microscope for inspecting electrical faults in integrated
circuits. Opt. Lett., 28(21):2058–2060, Nov 2003.
[4] A J Fitzgerald, E Berry, N N Zinovev, G C Walker, M A Smith, and J M Cham-
berlain. An introduction to medical imaging with coherent terahertz frequency
radiation. Physics in Medicine and Biology, 47(7):R67, mar 2002.
[5] Bradley Ferguson and Xi-Cheng Zhang. Materials for terahertz science and tech-
nology. Nature Materials, 1(1):26–33, Sep 2002.
[6] A. Giles Davies, Andrew D. Burnett, Wenhui Fan, Edmund H. Linfield, and John E.
Cunningham. Terahertz spectroscopy of explosives and drugs. Materials Today,
11(3):18–26, 2008.
[7] Kodo Kawase, Yuichi Ogawa, Yuuki Watanabe, and Hiroyuki Inoue. Non-
destructive terahertz imaging of illicit drugs using spectral fingerprints. Opt. Ex-
press, 11(20):2549–2554, Oct 2003.
[8] Marco F. Duarte, Mark A. Davenport, Dharmpal Takhar, Jason N. Laska, Ting
Sun, Kevin F. Kelly, and Richard G. Baraniuk. Single-pixel imaging via compressive
sampling. IEEE Signal Processing Magazine, 25(2):83–91, 2008.
[9] C.E. Shannon. Communication in the presence of noise. Proceedings of the IRE,
37(1):10–21, 1949.
[10] D.L. Donoho. Compressed sensing. IEEE Transactions on Information Theory,
52(4):1289–1306, 2006.
[11] E.J. Candes, J. Romberg, and T. Tao. Robust uncertainty principles: exact signal
reconstruction from highly incomplete frequency information. IEEE Transactions
on Information Theory, 52(2):489–509, 2006.
[12] Richard G. Baraniuk. Compressive sensing [lecture notes]. IEEE Signal Processing
Magazine, 24(4):118–121, 2007.
[13] Shaobing Chen and D. Donoho. Basis pursuit. In Proceedings of 1994 28th Asilomar
Conference on Signals, Systems and Computers, volume 1, pages 41–44 vol.1, 1994.
[14] Joel A. Tropp and Anna C. Gilbert. Signal recovery from random measurements
via orthogonal matching pursuit. IEEE Transactions on Information Theory,
53(12):4655–4666, 2007.
[15] Jian Wang, Seokbeop Kwon, and Byonghyo Shim. Generalized orthogonal matching
pursuit. IEEE Transactions on Signal Processing, 60(12):6202–6216, 2012.
[16] Marco F. Duarte and Richard G. Baraniuk. Kronecker compressive sensing. IEEE
Transactions on Image Processing, 21(2):494–504, 2012.
[17] Wei-Chieh Wang, Jou-Chen Wang, Yi-Chun Hung, Shang-Hua Yang, and Yuan-
Hao Huang. Tensor-based terahertz compressive sensing imaging. In 2021 46th
International Conference on Infrared, Millimeter and Terahertz Waves (IRMMW-
THz), pages 1–2, 2021.
[18] Qun Li, Dan Schonfeld, and Shmuel Friedland. Generalized tensor compressive
sensing. In 2013 IEEE International Conference on Multimedia and Expo (ICME),
pages 1–6, 2013.
[19] Cesar F. Caiafa and Andrzej Cichocki. Computing sparse representations of mul-
tidimensional signals using kronecker bases. Neural Computation, 25(1):186–220,
2013.
[20] Wei-Chieh Wang, Yi-Chun Hung, Yu-Heng Du, Shang-Hua Yang, and Yuan-Hao
Huang. Fpga-based tensor compressive sensing reconstruction processor for tera-
hertz single-pixel imaging systems. IEEE Open Journal of Circuits and Systems,
3:336–350, 2022.
[21] Jens Neu and Charles A. Schmuttenmaer. Tutorial: An introduction to terahertz
time domain spectroscopy (THz-TDS). Journal of Applied Physics, 124(23):231101,
12 2018.
[22] L. Duvillaret, F. Garet, and J.-L. Coutaz. A reliable method for extraction of ma-
terial parameters in terahertz time-domain spectroscopy. IEEE Journal of Selected
Topics in Quantum Electronics, 2(3):739–746, 1996.
[23] Anton Travinsky, Dmitry Vorobiev, Zoran Ninkov, Alan Raisanen, Manuel A. Qui-
jada, Stephen A. Smee, Jonathan A. Pellish, Tim Schwartz, Massimo Robberto,
Sara Heap, Devin Conley, Carlos Benavides, Nicholas Garcia, Zach Bredl, and Se-
bastian Yllanes. Evaluation of digital micromirror devices for use in space-based
multiobject spectrometer application. Journal of Astronomical Telescopes, Instru-
ments, and Systems, 3(3):035003, 2017.
[24] Szu-Hsi Chen, Chia-Ming Mai, Yi-Chun Hung, Shang-Hua Yang, and Yuan-Hao
Huang. 3d tensor compressive sensing thz single-pixel imaging for refractive index
estimation. In 2023 48th International Conference on Infrared, Millimeter, and
Terahertz Waves (IRMMW-THz), pages 1–2, 2023.
[25] I. Daubechies, M. Defrise, and C. De Mol. An iterative thresholding algorithm for
linear inverse problems with a sparsity constraint. Communications on Pure and
Applied Mathematics, 57(11):1413–1457, 2004.
[26] David L. Donoho, Arian Maleki, and Andrea Montanari. Message-passing algo-
rithms for compressed sensing. Proceedings of the National Academy of Sciences,
106(45):18914–18919, 2009.
[27] Rayko Ivanov Stantchev, Xiao Yu, Thierry Blu, and Emma Pickwell-MacPherson.
Real-time terahertz imaging with a single-pixel detector. Nature communications,
11(1):2535, 2020.
[28] Zhou Wang, A.C. Bovik, H.R. Sheikh, and E.P. Simoncelli. Image quality assess-
ment: from error visibility to structural similarity. IEEE Transactions on Image
Processing, 13(4):600–612, 2004.
[29] Scott J. Campbell and Sunil P. Khatri. Resource and delay efficient matrix mul-
tiplication using newer fpga devices. In Proceedings of the 16th ACM Great Lakes
Symposium on VLSI, GLSVLSI ’06, page 308–311, New York, NY, USA, 2006.
Association for Computing Machinery.
[30] Hsu-Hang Yu. Two-stage compressive sensing and approximate message passing
signal reconstruction processor for terahertz single-pixel compressive sensing imag-
ing systems. Master’s thesis, National Tsing Hua University, Taiwan, 2023.
[31] Jienan Chen, Zhenbing Zhang, Hao Lu, Jianhao Hu, and Gerald E. Sobelman. An
intra-iterative interference cancellation detector for large-scale mimo communica-
tions based on convex optimization. IEEE Transactions on Circuits and Systems
I: Regular Papers, 63(11):2062–2072, 2016.
[32] Xilinx. UltraScale Architecture Configurable Logic Block User Guide (UG574). Xil-
inx.
[33] Jun Li, Paul Chow, Yuanxi Peng, and Tian Jiang. Fpga implementation of an
improved omp for compressive sensing reconstruction. IEEE Transactions on Very
Large Scale Integration (VLSI) Systems, 29(2):259–272, 2021.
[34] Hassan Rabah, Abbes Amira, Basant Kumar Mohanty, Somaya Almaadeed, and
Pramod Kumar Meher. Fpga implementation of orthogonal matching pursuit for
compressive sensing reconstruction. IEEE Transactions on Very Large Scale Inte-
gration (VLSI) Systems, 23(10):2209–2220, 2015.
[35] Xiang Ge, Fan Yang, Hengliang Zhu, Xuan Zeng, and Dian Zhou. An efficient
fpga implementation of orthogonal matching pursuit with square-root-free qr de-
composition. IEEE Transactions on Very Large Scale Integration (VLSI) Systems,
27(3):611–623, 2019.
[36] Thomas James Thomas and J. Sheeba Rani. Fpga implementation of sparsity
independent regularized pursuit for fast cs reconstruction. IEEE Transactions on
Circuits and Systems I: Regular Papers, 69(4):1617–1628, 2022.