Ultrasonic material characterization using Matching Pursuit algorithm with experimental dictionary

Adaptation of the Matching Pursuit algorithm is presented to simulate the successive overlapped echoes of an ultrasonic signal. Crucial step of sparse representation algorithm such as MP algorithm is the choice of the dictionary. Here, the dictionary is built from experimental measurements in a transmission technique. Acoustic measurements on copper and ABS sample at 500 kHz, 10 MHz, 15 MHz and 20 MHz are performed to test the efficiency of the method. Simultaneously measurements of thickness and velocity of the copper sample are reported at different frequency. These results show that using a dictionary construct from the reference signal is enough to obtain sample characteristics with low attenuating material. The robustness of the method is studied as a function of time delay between the different echoes. In the case of the ABS sample, taking into account the attenuation of this material in the development of the dictionary is critical to identifying a round-trip signal when overlapping occurs.