Signal Management With Median

For measures recorded from the colour reader, the ultrasonic distance sensor and other sophisticated devices prone to erratic values, we need to separated the signal from the noise.

By signal, we mean usable information. At the contrary, noise refers to contaminated and useless measures.

Let's consider two common solutions

average
median

Let's take this example with 5 measures

At first sight, there's something odd with the 2nd measure, 3600.

Average

The average is the sum of all measures divided by the number of measures.

In this case, average is 2465.

But this result takes into account the odd measure of 3600.

Without that 3600 odd measure, result would have been 2181, more consistent with the 4 measures.

So we need another method.

Median

As per wikipedia,

In probability theory and statistics, a median is described as the numeric value separating the higher half of a sample, a population, or a probability distribution, from the lower half.

So, if we have five measures, we're going

first to sort them
and then take the value in to middle (i.e. the 3rd one).

Result is 2193.

Ready-To-Use Shared Function

Please find below the Robo Pro project ready-to-use Median functions.

Three options:

Median based on 3 measures, to be used in standard situation
Median based on 5 measures, to be used in difficult situation
Median based on 7 measures, to be used in critical situation

Keep in mind:

The more noise prone measure, the higher the number of measures needed,
The higher the number of measures, the slower the function.

Download

Download the driver with an example for RoboPro.

This file is under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

mediane.rpp.zip
File Size:	29 kb
File Type:	zip

Download File