Application of variant least mean squares adaptive algorithms for filtering material grain noise

Some least mean squares (LMS)-based adaptive filters have been used efficiently for noise reduction in ultrasonic testing (UT). However, the realisation of these filters in UT systems is hindered by the limited computational resources of the hardware. This work proposes to apply simplified variants of the LMS adaptive filter for filtering the grain noise in UT. These variants, the sign-error, sign-data and sign-sign LMS adaptive filters, simplify the conventional LMS adaptive filter and facilitate its implementation in hardware. The sign LMS adaptive filters are evaluated by experiments comparing their performance in filtering grain noise with that of the conventional LMS. The results show that the signdata LMS filter is a good alternative to the conventional LMS adaptive filter in this application.