Ultrasonic Image Artifacts Removing Technique Based on SoS Kernel

Abstract

The ultrasonic image provides direct visualization information of the test sample. However, traditional ultrasonic images often accompany artifacts inevitably, which heavily influence the accurate quantization and evaluation of defects. So far, it has been hard to find a way for eliminating artifacts caused by the diffusion of isacoustic contour, signal tail, or noise simultaneously. To remove different kinds of ultrasonic image artifacts, a novel ultrasonic imaging method is presented using a SoS (Sum of Sinc) kernel. If the envelope of ultrasonic echo is obtained, let it pass through a SoS kernel, then sample the signal at equal intervals determined by the echo signal information degree, and apply the Fourier transform to the discrete sampling data to obtain the Fourier coefficient sequence. After that, use the spectral estimation algorithm to estimate the parameters of the ultrasonic echo signal and reconstruct the echo signal using the estimated parameters. Finally, the ultrasonic image can be obtained by calculating the acoustic field using the reconstructed echoes. Experiments have been performed on a sample with standard defects of straight slot and through-hole, and the experimental results revealed that the imaging method not only improves the quantization accuracy of the defect but also greatly restrains the different types of artifacts.