Simulation of THz-TDS echo filtering algorithm based on deep learning

Hongfei Zhang; Yuanmeng Zhao; Cunlin Zhang

doi:10.1117/12.2573406

10 October 2020 Simulation of THz-TDS echo filtering algorithm based on deep learning

Hongfei Zhang, Yuanmeng Zhao, Cunlin Zhang

Proceedings Volume 11559, Infrared, Millimeter-Wave, and Terahertz Technologies VII; 115590O (2020) https://doi.org/10.1117/12.2573406
Event: SPIE/COS Photonics Asia, 2020, Online Only

Conference Poster

Abstract

We analyzed the reasons why ZnTe generates terahertz echo. Based on the Gaussian beam, the terahertz signal with echo is produced by simulation. In this paper, the deconvolution algorithm is further improved. The linear assumption that sample absorption is proportional to frequency is abandoned. We studied the method of echo based on Echo State Network. Set 200 neural units in each layer of the network. We specify 250 rounds of training. To prevent the gradient from exploding, set the gradient threshold to 1. Specify an initial learning rate of 0.005. After 125 rounds of training, multiply it by a factor of 0.2 to reduce the learning rate. This method largely removes echo. But it has not been completely removed, and there is still less than 5% of the echo signal remaining. We consider the reason may be that the number of neurons, the number of iterations, weight setting and other parameters have yet to be optimized. Although this method does not completely remove the echo, but provides a new method for removing the echo. By optimizing parameters such as the number of neurons, the number of iterations, weight setting and increasing the amount of training data, this method is expected to become a new method for echo removal.

Conference Presentation

Citation Download Citation

Hongfei Zhang, Yuanmeng Zhao, and Cunlin Zhang "Simulation of THz-TDS echo filtering algorithm based on deep learning", Proc. SPIE 11559, Infrared, Millimeter-Wave, and Terahertz Technologies VII, 115590O (10 October 2020); https://doi.org/10.1117/12.2573406

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available