TR2021-098

Learning-Based THz Multi-Layer Imaging for High-Capacity Positioning


    •  Wang, P., Koike-Akino, T., Ma, R., Orlik, P.V., Yamashita, G., Tsujita, W., Nakajima, M., "Learning-Based THz Multi-Layer Imaging for High-Capacity Positioning", International Conference on Infrared, Millimeter, and Terahertz Waves (IRMMW-THz), November 2021.
      BibTeX TR2021-098 PDF
      • @inproceedings{Wang2021nov,
      • author = {Wang, Perry and Koike-Akino, Toshiaki and Ma, Rui and Orlik, Philip V. and Yamashita, Genki and Tsujita, Wataru and Nakajima, M.},
      • title = {Learning-Based THz Multi-Layer Imaging for High-Capacity Positioning},
      • booktitle = {International Conference on Infrared, Millimeter, and Terahertz Waves (IRMMW-THz)},
      • year = 2021,
      • month = nov,
      • url = {https://www.merl.com/publications/TR2021-098}
      • }
  • MERL Contacts:
  • Research Areas:

    Computational Sensing, Machine Learning, Optimization, Signal Processing

Abstract:

This paper demonstrates a learning-based THz multi-layer pixel identification for contactless three-dimensional (3-D) positioning and encoders. More specifically, we propose a one-dimensional convolution-based residual network to deal with practical issues including 1) depth variation, 2) shadow effect, and 3) content recognition at the back surface of each layer. Experimental validation on a three-layer sample with contents on all surfaces is also provided.