A rapid approach to speckle noise reduction in ultrasonic non-destructive evaluation using matched filters

Abstract

In this paper, we present a rapid technique to reduce the speckle noise in ultrasonic non-destructive evaluation (NDE) and enhance the image quality using matched filters. This method implicitly explores the signatures and profiles of echoes from legitimate defects and echoes from random reflectors, uses the temporal and spectral distinction to design the filter, and utilizes a particle swarm optimization (PSO) paradigm to optimize the filter parameters with an objective to maximize the output signal-to-noise ratio (SNR). An optimized matched filter approximating the desired defect echoes is then applied to the received A-scan waveforms. Experiments with a 128-element 5MHz transducer on a mild steel sample are conducted. It has been demonstrated that clutter is significantly reduced and the image SNR is improved by more than 20dB when the matched filter is applied to A-scan waveforms prior to image formation. Given the fact that the matched filter can be implemented in real-time, the great performance advantages are obtained with extremely low extra computational cost.