US 11,719,796 B2
System and method for ultrasonic signal noise removal using a deep neural network
Ying Li, Sunnyvale, CA (US); Usha Nookala, Sunnyvale, CA (US); and Sihao Ding, Sunnyvale, CA (US)
Assigned to Volvo Car Corporation, Gothenburg (SE)
Filed by Volvo Car Corporation, Gothenburg (SE)
Filed on Dec. 31, 2020, as Appl. No. 17/138,984.
Prior Publication US 2022/0206130 A1, Jun. 30, 2022
Int. Cl. G01S 7/52 (2006.01); G01S 15/10 (2006.01); G06N 3/08 (2023.01); G01S 7/539 (2006.01); G06F 18/2132 (2023.01)
CPC G01S 7/52077 (2013.01) [G01S 7/539 (2013.01); G01S 15/101 (2013.01); G01S 15/104 (2013.01); G06F 18/2132 (2023.01); G06N 3/08 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method, comprising:
receiving an input ultrasonic signal at a generator module of a generative adversarial network; and
removing noise from the ultrasonic signal using the generator module of the generative adversarial network to generate an output denoised ultrasonic signal;
wherein the input ultrasonic signal comprises a one-dimensional input ultrasonic signal; and
wherein the one-dimensional input ultrasonic signal is pre-processed before being received at the generator module of the generative adversarial network by converting the one-dimensional input ultrasonic signal into a two-dimensional M×N array that replaces the input ultrasonic signal that is received at the generator module of the generative adversarial network.