A Deep Learning Approach for Artifact Suppression in MEMS-based Airborne Ultrasonic Transceivers

In this study, we introduce a denoising autoencoder as a solution to mitigate artifacts in MEMS-based airborne ultrasound transceivers. The autoencoder employs a deep neural network architecture to learn a robust representation of ultrasonic echo signals produced by reflecting targets, effectively eliminating unwanted noise. Experimental pulsed-echo ultrasonic recordings are used both in the development and evaluation of the model. Our results highlight the denoising autoencoder’s ability to isolate echoes, even in complex scenarios where noise artifacts are prevalent. Evaluation based on the Mean Square Error (MSE) criterion reaffirms the success of the proposed method in effectively removing noise artifacts without suppressing the essential echo signals, as evidenced by the low MSE value of 7.89 • 10
^-4
.