Noise Removal

FFT Engine

Noise Removal

The FFT Engine contains five different noise removal processes

**Click Removal

Median Filter

Average Subtraction

Gaussian Kernel Smoothing

Thresholding**

Each method can be enabled individually

Click Removal

The click removal method operates on the time series data prior to the FFT calculation and therefore affects both output streams of the FFT Engine.

Click removal measures the standard deviation of the time series data and then multiplies the signal by a factor which increases rapidly for large signal components. This has the effect of reducing the magnitude of short duration transient signals such as echolocation clicks

Median Filter

Within each spectrogram slice, the median value about each point is taken and subtracted from that point.

Average Subtraction

A decaying average spectrogram is computed and subtracted from the current spectrogram value.

Gaussian Kernel Smoothing

The spectrogram is smoothed by convolving the image with a Gaussian smoothing kernel

1 2 1
2 4 2
1 2 1.

Thresholding

A threshold is applied and all data falling below that threshold set to 0.

Previous: Creating an FFT Engine