Abstract:In the actual onedimensional signal processing, the mathematical morphological filtering method (MMFM) has a large amount of operation. To address this issue, a fast mathematical morphological filtering method (FMMFM) is proposed in this study. First, the erosion operation is modified to realize the realtime signal processing. Then, the data updating process of buffer in the microprocessor system is combined with the theory of MMFM. The iteration of sliding window is used to improve the time consumption of MMFM. In further, the smooth filtering method is also improved to speed up the optimization of the FMMFAM results. The measured pulse signals are used as experimental data. Compared with the MMFM, results show that the FMMFM can effectively reduce the calculation time (speed up over 70 times) and keep the filtering accuracy unchanged. The FMMFM with flat and linear structuring elements have faster filtering speed than those of other elements (speed up over 110 times). The proposed method can still process the signal in realtime (less than 45 s for the signals of 240 s) as the increase of the lengths of structuring element and buffer. It can be employed in pulse signal filtering, segmenting and feature extracting in realtime (less than 45 s). Therefore, the proposed method may be applied in some smart wearable devices with high realtime requirements, such as wristbands and watches.