Block-matching and 3D filtering-based denoising of acoustic images obtained through point contact excitation and detection method

Abstract

A novel experimental method based on the Coulomb coupling technique is used for visualizing the interaction of ultrasonic waves in PZT (Lead Zirconate Titanate - (Pb[ZrxTi1-x]O3 (0 ≤ x ≤ 1)) ceramic materials. The method of excitation and detection through a point contact produces representations of propagating wave vectors in the time domain as a series of images. In the acoustic image, several types of noises from different types of sources corrupt the signal as well as the image. As a result, the resolution and quality of the images are reduced. Block matching and 3D filtering (BM3D) algorithm is implemented for the reduction of noises from ultrasonic signals that are in the form of 2D images. Here, we also implemented a custom-designed hybrid filter and conventional total variation (TV-L1) filter on the noisy images to compare the final denoised outputs from the BM3D filter. Comparison metrics like PSNR (peak signal-to-noise ratio) and SSIM (structural similarity index measure) are used for quantitative analysis of those denoised ultrasonic images with respect to their ground truth or almost noise-free counterparts. Based upon those parameter values, the BM3D filter demonstrated the best results for the denoising of the acoustic images acquired from the sample PZT material. This paper discusses and explains the results and proves the successful application of the BM3D algorithm on acoustic image data.