Imaging in the presence of grain noise using the decomposition of the time reversal operator.

In this paper, we are interested in detecting and imaging defects in samples of cylindrical geometry with large speckle noise due to the microstructure. The time reversal process is an appropriate technique for detecting flaws in such heterogeneous media as titanium billets. Furthermore, time reversal can be iterated to select the defect with the strongest reflectivity and to reduce the contribution of speckle noise. The DORT (the French acronym for Decomposition of the Time Reversal Operator) method derives from the mathematical analysis of the time reversal process. This detection technique allows the determination of a set of signals to be applied to the transducers in order to focus on each defect separately. In this paper, we compare three immersion techniques on a titanium sample, standard transmit/receive focusing, the time reversal mirror (TRM), and the DORT method. We compare the sensitivity of these three techniques, especially the sensitivity to a poor alignment of the array with the front face of the sample. Then we show how images of the sample can be obtained with the TRM and the DORT method using backpropagation algorithm.