Lowpass filters are used to discriminate slowly changing features (low frequencies) from rapidly changing features in data (high frequencies). The application of a lowpass filter results in a new series consisting of the smooth features in the original data. In situations where noise or external artifacts are known to contaminate data in high frequencies, for example, the application of a lowpass filter is desirable.

Subtracting the result of a lowpass filter from the original data yields the high frequency components of the data. This technique is often used to separate a signal, whose parts are then explored independently.

Results of Applying a Lowpass Filter

Here the result of the filter was the low frequency signal shown in green in the graph. The high frequency signal (blue) was obtained by subtracting the low frequency signal from the original signal (red).

A highpass filter is similar to a low pass filter in principle, but extracts rapidly changing features as opposed to the slowly changing features of data.

A bandpass filter will allow some intermediate frequencies to pass, as specified by the low and high cutoff frequencies. A bandstop filter does the opposite of a bandpass filter—it prevents some intermediate frequencies to pass.