Wavelet denoising applied to light emitting diode lidar signal

• Main Authors: Ong, Prane Mariel, Galvez, Maria Cecilia, Vallar, Edgar C., Shiina, Tatsuo
• Format: text
• Published: Animo Repository 2017
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/830; https://animorepository.dlsu.edu.ph/context/faculty_research/article/1829/type/native/viewcontent
• Institution: De La Salle University

© 2017 American Scientific Publishers. All rights reserved. LED lidar is relatively new in the field of remote sensing. Although, it has a higher tendency to be embedded in noise than the conventional laser lidar due to its relatively low backscattered signal, the potential of portability is promising. Thus, in this study, established denoising technique conventionally applied to laser lidar backscattered signal was applied in LED Lidar backscattered signal, and its outcome was investigated. LED lidar backscattered signals were denoised using 9 preselected wavelet functions: db3, db5, db7, coif3, coif5, sym3, sym5, sym7, and haar, and 6 thresholding parameters. The result showed that ‘heursure,' soft thresholding, with the ‘sln’ scaling parameter were the best thresholding parameter combination since they gained the highest signal-to-noise ratio (SNR). It also showed that regardless of the wavelet variants used with respect to the best thresholding parameter there was no statistically significant difference in their SNR values. Nevertheless, coif5 was used as the representative wavelet to demonstrate the comparison of the original LED Lidar backscattered signal with the wavelet denoised signal, and with the range-squared corrected signal, since it got the highest SNR. By visual inspection, some speckled noise from the original LED Lidar signal were smoothened.