As is often the case, experimental data contain various superimposed signals, from which the required signal or signals must be extracted. In such cases, the required signal level has not always been higher than the other signal levels. Therefore, it was difficult to eliminate all unnecessary signals using conventional noise elimination methods by classifying unnecessary signals as noise. To overcome such difficulties, a multistage signal extraction method is suggested which extracts the desired signal component by applying the filtering in multi-stages. The results extracted by this method contain only the required signals plus the noise with the same amplitude readings. This makes it easier to reduce the complexity of signal extraction in later stages, such as the signal extraction based on other data information.

In this study, a spline filter is constructed using the cubic B-spline smoothing process, which has versatility and is employed widely as a practical method. The multistage signal extraction method has versatility, and filters other than the spline filter can be utilized if the waveform is not overly distorted.