Im creating an application where i need to position a ImageView depending on the Orientation of the device. I use the values from a MagneticField and Accelerometer Sensors
A low pass filter (LPF) blocks fast changing signals and
allows only slow changes in the signals. This means any small
sudden changes will be ignored.
The standard way to implement this in software is to take a running average
of the last N samples and report that value. Start with N as small as 3 and
keep increasing N until you find sufficient smoothed out response in your app.
Do keep in mind that the higher you make N, slower the response of the system.